Nederlandstalige samenvatting. 97. Appendix. 103 ...... worden dat de PPC anatomische submodules heeft ... fysiologie en haar relatie tot bevindingen in niet-.
Cortical oscillatory activity in human visuomotor integration Een wetenschappelijke proeve op het gebied van de sociale wetenschappen
Proefschrift
ter verkrijging van de graad van doctor aan de Radboud Universiteit Nijmegen op gezag van de rector magnificus prof. mr. S.C.J.J. Kortmann, volgens besluit van het College van Decanen in het openbaar te verdedigen op 30 maart 2011 om 13.30 uur precies
door
Jurrian van der Werf geboren op 07 april 1982 te Losser
1
Promotor: Co-promotor:
prof. dr. H. Bekkering dr. W.P. Medendorp
Manuscriptcommissie:
prof. dr. D.F. Stegeman (voorzitter) prof. dr. C.C.A.M. Gielen prof. dr. P.R. Roelfsema (Vrije Universiteit, Amsterdam)
ISBN 978-90-9026052-5 The research presented in this research was financially supported by the Netherlands Organization for Scientific Research (NWO). Printed by Ipskamp Drukkers, Enschede, The Netherlands. Lay-out inspired by dr. G. van Elswijk’s thesis, 2008. © Jurrian van der Werf, 2011 2
“I’m pickin’ up good vibrations She’s giving me excitations” Brian Wilson/Mike Love
3
4
Contents 1. General introduction
9
2. Gamma-band activity in human posterior parietal cortex encodes the motor goal during delayed prosaccades and antisaccades
29
3. Neuronal synchronization in human parietal cortex during saccade planning
45
4. Reorganization of oscillatory activity in human parietal cortex during spatial updating
55
5. Neuronal synchronization in human posterior parietal cortex during reach planning
73
6. Summary
91
7. Nederlandstalige samenvatting
97
Appendix
103
5
Voorwoord
Voordat je van de betere onderdelen van dit proefschrift gaat genieten, bedenk dat deze alleen tot stand hebben kunnen komen door de hulp en steun van een enorm aantal mensen! Als eerste wil ik Pieter Medendorp noemen. Natuurlijk omdat hij me de kans heeft gegeven aan dit project te beginnen, maar ook voor de vorm en inhoud van z’n begeleiding; door genoeg ruimte te geven, te ondersteunen waar nodig en aan te jagen als het allemaal wel weer lang genoeg had geduurd heeft hij dit proefschrift en mij in goede staat door vier+ tumultueuze jaren geloodst. Dat hij altijd en overal geïnteresseerd is in de laatste ontwikkelingen in het project is bovenal een uiting van zijn betrokkenheid; iets dat ik altijd zeer heb weten te waarderen. Wie kan bovendien zeggen dat zijn begeleider Albert Heijn voetbalplaatjes van z’n zoontje steelt om z’n promovendus aan het logo van FC Twente te helpen? Om deze en meerdere redenen kan ik zeggen: Pieter, mijn dank is groot in vele referentiekaders! Ook ben ik veel dank verschuldigd aan mijn sensorimotorlabmaatjes, in order of appearance, Rens (bedankt voor de vacaturetip!), Sabine, Stan, Maaike, Verena (bedankt voor je bijdrage aan hoofdstuk 3 en vooral 4!), Frank, Ivar, Tobias, Joost en Luc. De sfeer was niet alleen op de afdeling altijd goed, maar ook daarbuiten, en dan met name de jaarlijkse labuitjes naar de VS zijn onvergetelijk. Als tweede wil ik Ole Jensen bedanken. Zonder de wekelijkse meetings met Ole hadden de analyses een stuk langer geduurd en waren de 4 jaren een stuk minder gezellig geweest. Ole, bedankt! Ook Ole’s groep was belangrijk, want altijd was er wel iemand behulpzaam én vrolijk! Bedankt Hanneke, Ali M., Dasha, Ingrid, Daniel, Christian (“spaceman”), Barbara, Marcel, Ali B., Jan, Saskia, Freek, Stephen, Niels, Yuka, Paul, Mathilde en Avgis voor alle hulp, 6
gezelligheid en input bij meetings en daarbuiten. Ook wil ik Pascal Fries bedanken voor z’n bijdrage aan onze gezamelijke projecten. Een meeting met Pascal stond garant voor enkele weken extra analyses maar ook voor dramatische verbeteringen van manuscripten en kennis. Harold Bekkering, al was hij niet direct betrokken bij de dagelijkse gang van zaken, ik hou hem toch verantwoordelijk voor de prettige sfeer en strakke organisatie op het NICI (en DCC). Dank ook aan alle NICI mensen en later DCC mensen, die de koffiepauze tot het hoogtepunt van de werkdag maakten. In een adem door wil ik ook het management en secretariaat op het DCC bedanken: Beppie, Yvonne, Saskia, Maaike en Johanna, bedankt voor alles wat jullie zichtbaar en onzichtbaar voor me hebben gedaan. Hetzelfde geldt voor het management en de onmisbare technische ondersteuning aan de kapittelweg. Bedankt Arthur, Tildie, Sandra en Nicole, Erik, Bram, Marek, Sander, Paul G., René, Rene en Edward. De kleurenplaatjes in dit proefschrift hadden er zonder overdrijven zonder FieldTrip heel anders uitgezien. Enorm veel dank hiervoor gaan naar Dr. Robert (“... he helps you to understand/he does everything he can/dr Robert” (Lennon and McCartney, 1966)) en de rest van de FieldTrippers, met name Jan-Mathijs, Ingrid (sourceplot.m was een gamechanger!) en Eric Maris. Minstens zo belangrijk voor de mooie figuren waren de hersenen waar de figuren van afgeleid zijn, dus de dappere vrijwilligers die elke keer weer geheel belangeloos uren van hun tijd opofferden om in het donker eindeloos veel oogbewegingen te maken. Sander verdient hier een speciale vermelding voor z’n deelname aan ál mijn hier beschreven (maar ook de niet-beschreven) experimenten, en z’n bijna-perfecte alpha lareralisatie! Voor m’n paranimfen, Anil en Stan, op voor-
hand dank dat jullie zo onzettend relaxt gaan zijn op 30 maart en dat jullie dat eigenlijk altijd zijn geweest ook. Als ik bij een pittige vraag niet direct in de foetushouding kruip is dat door jullie lage hartslag. Kamergenoten; hoe overleef je 4 jaar promotieonderzoek als je niet eens in de zoveel tijd lief en leed kan delen in de kamer waar je zoveel tijd doorbrengt? Shaozheng, Judith, Sabine, Ivar en Matthias, bedankt voor het er zijn. In reactie op de overdadige dankbetuiging van de “Fries” op nummer 50: geen dank. Verder wil ik de auteurs van Balvers et al. 2010; Tuladhar et al., 2010; Aarntzen et al., 2008; Batmaz, 2003 en Peters et al., 2011 bedanken voor het lage wetenschappelijke gehalte van
onze gesprekken en bezigheden. Die dank geldt natuurlijk met net zoveel plezier voor Qiang, Laura, Silvie, Arno, Sabrina, Anke, Nicole, Vincent, Marc, Reinier, Lisette, Marleen, Anthony, Johan, Ivar, Guen, en iedereen die me nu even ontschoten is die me een leven buiten het lab en de wetenschap heeft gegeven! Ook dank voor de onvoorwaardelijke steun en vertrouwen dat ik écht wel iets aan het doen was in Nijmegen de afgelopen 5 jaar (zie hier waar het toe geleid heeft...) aan m’n moeder, vader, Jutta en de kids, Willem (check de achterkant van het proefschrift; speciaal voor jou!), Shalaka, Els en Henri. En de belangrijkste plek is voor jou, Esther, voor al onze good vibrations en excitations! Jij maakt me écht gelukkig. Synch! :o)
7
8
Chapter 1 General introduction
9
Chapter 1
H 1
ow does our brain represent space? And how does it use this representation for action? These are the two driving questions behind this thesis. Space, as I define it here, is the boundless, three-dimensional extent in which objects have relative positions. In the context of this thesis, these objects have a potential behavioral relevance to us, meaning that they can influence or guide our actions. One such an object could be a cup of coffee in front of us, or a pencil on the table that we want to act upon by picking it up. Further away, a bright star shining in the night can also be such an object, that we may want to act upon by directing our gaze toward it. To construct a spatial representation, the brain depends on inputs from our senses. We can see, hear or touch objects, and use all of this information to infer their positions relative to ourselves and one another. The sense of vision is most dominant in our perception of space. We use our visual perceptions in all kind of daily routines like reading, communication and safely moving around. Despite these complex cognitive operations that are associated with vision in the modern human, it is thought that visual perception once developed as a means to guide simple movements in our peripersonal space. Small groups of light sensitive cells allowed early invertebrate animals to vaguely observe objects in their near surrounding, enabling them to move toward them, or away, which gave these animals a slight advantage in the game of adaptation and survival. With the further development of the eyes and limbs, early primates gained more advantage by carefully coordinating their eye and limb movements, allowing for more complex goal-directed motor acts and eventually even tooluse. In the modern primate brain, this goal-directed processing of vision is expressed in a distinct visual pathway, known as the dorsal visual pathway (Ungerleider and Mishkin, 1982). Information from the retina is passed on to the occipital cortex, also 10
known as the striate visual cortex. From here, the vision-derived information flows through various extrastriate areas, terminating in the posterior parietal cortex (PPC). The PPC receives many and various inputs; not only visual, but also auditory, somatosensory, limbic and motor output signals. The PPC is interconnected with the motor areas in the frontal cortex, which are involved in generating motor commands to control the muscles. This sounds all simple enough, but our brain needs to resolve several demanding spatial transformations to arrive at the high level of visuomotor behavior that characterizes primates. To understand the computational complexity of these processes, it is first necessary to introduce the lingua that neuroscientists use to describing space and spatial transformations. One element of this vocabulary is the concept of a reference frame, which, in the mathematical sense, is simply a set of rigid axes that intersect at a point, the origin. In the case of motor control, the more stable insertion point of a set of muscles is typically taken as the frame of reference, like the head for eye movements, and the torso for head and arm movements. The coordinate axes within such a frame can be taken perpendicular to each other (Cartesian coordinates), or can be defined in terms of a radius and direction (polar coordinates), and they are marked with gradations. This system allows the location of any object to be described by a set of numbers, called coordinates – its position along each of the axes. With these concepts at hand, one could formulate questions about how the brain encodes space and actions in very concrete terms. For example, suppose we wish to understand the simple task of picking up a cup of coffee. Using these concepts, understanding the underlying visuomotor transformation becomes a matter of providing an account of how the brain translates the position of the cup from the coordinates of the retinas into the coordinates of a reference frame that links to the hand (see Figure 1.1).
General introduction
Figure 1.1 Schematic representation of the two possible vision-to-reach reference frames. The red vectors together make up the gaze-centered reference frame. Targets are encoded relative to the gaze, as is the position of the hand. By subtracting the hand vector from the target vector the movement vector is calculated. The blue vectors make up the body-centered or shoulder-centered representation. Both target and hand are encoded relative to the shoulder. A similar computation as in the gaze-centered reference frame results in the movement vector, now calculated in a body-centered reference frame. T, target position; H, hand position; M, movement vector; B, body-centered coordinates; E, gazecentered coordinates. (Adapted From Buneo and Andersen, 2006).
How information is transformed across the various reference frames is still not fully understood and one of the core issues of this thesis. However, we do know that the posterior parietal cortex plays a crucial role in this process of sensorimotor integration (Snyder et al., 1997; Buneo et al., 1999; Pesaran et al., 2006). In this thesis, we present research providing new insights into the parietal mechanisms underlying the visuomotor transformations and spatial representations for actions. We have assessed neuronal activity in the sensorimotor system by measuring temporally synchronized neuronal activity (also called neuronal oscillations) in humans with magnetoencephalography (MEG), in order to provide novel knowledge about the temporal dynamics of sensorimotor trans-
formations. With this technique, we have probed questions concerning the underlying neuronal dynamics of movement planning in parietal cortex, as well as the nature of their associated spatial representations. Also, we have established the reference frames in which the neuronal oscillations represent action representations. Finally, we have looked at the role of neuronal synchronization when coding spatial representations for different effectors. In this general introduction, I will first present some background on the PPC’s functional organization subserving sensorimotor transformations, particularly visuomotor integration. Then, I will present some background on the MEG and neuronal oscillations, and finally I will formulate the research questions that are addressed in this thesis in more detail. Posterior Parietal Cortex Being anatomically positioned between the occipital and frontal cortex, thus functionally between perception and action, the parietal cortex would seem like the ideal candidate to perform visuomotor transformations (see Figure 1.2). Indeed, patients with damage to the parietal cortex are unaffected in their primary visual and motor functions, but are affected when they try to connect the two. For example, such patients could make large errors when reaching to visual objects in their visual periphery (Goodale and Milner,1992), a condition called optic ataxia. Commonly, but not always, these patients show spatial neglect, a symptom in which the subject is unable to perceive space contralateral to the affected cortical hemisphere. This means that the PPC is at some level involved in representing space, and one that is important for subsequent action. Despite the important insights obtained from neuropsychological studies, they may not allow for strong conclusions about the functions of specific brain areas. Lesions are rarely confined to only one specific brain structure, and stroke patients more 11
1
Chapter 1
1
often than once present a multitude of function loss, not necessarily all directly attributable to the lesion. To better control the parameters influencing neuronal activity, neuroscientists resorted to different techniques, such as electrophysiological recordings in the monkey brain. The monkey brain, and particularly the rhesus macaque’s posterior parietal cortex (PPC), seems to be a good model for the human PPC, sharing a wide range of functions supporting comparable behavior (Grefkes and Fink, 2005). Figure 1.2 shows a human brain (left) and a macaque brain (right), both with the frontal cortex to the right, and the visual cortex (in grey) to the left. The parietal cortex is colored in blue. The darker blue represents the somatosensory area, the cortical site where tactile stimulation of the skin is processed, in a somatotopic fashion. The lighter shades of blue represent the different substructures within the PPC. In both species, the intraparietal sulcus (IPS) divides the PPC into a superior and inferior lobe. In the displayed human brain, the IPS is marked by a dashed line, while in the macaque brain it has been opened up to gain visual access to the different substructures within the IPS. Even
though there are large differences in the anatomy of the two different species (for instance, the human brain is twice the size of the macaque’s brain, and the human parietal surface is eight times as large as the macaque’s parietal cortex), most fundamental insights into the functions of the PPC have been gained by monkey electrophysiology and intervention studies. Therefore, I will start by discussing the anatomical and functional organization of the macaque’s PPC in relation to visuomotor processing, to later continue with the differences and similarities with the human PPC. Studies in awake and behaving macaque monkeys have identified several structures in the PPC, predominantly in and around the IPS, that are heavily involved in sensorimotor processing. Pragmatically named after their relative location in the IPS, we now know these regions as the anterior (AIP), lateral (LIP), medial (MIP), caudal (CIP) and ventral (VIP) intraparietal areas. Much research has gone into area LIP, which was recently found to be subdivided into a ventral and dorsal portion (LIPd and LIPv, respectively). Besides the IPS regions, also area 5 and the parieto-occipital sulcus (POS),
IPS A
B
PRR
Area 5
Figure 1.2 Human and macaque cortex. Posterior parietal cortex of human (left) and macaque monkey (right). A. The human parietal cortex can be divided into an anterior (dark blue), and posterior (light blue) part. The posterior parietal cortex (PPC) is divided by the intraparietal sulcus (IPS) into the superior parietal lobe (SPL) and the inferior parietal lobe (IPL). The IPL consists of the angular gyrus (Ang) and supramarginal gyrus (Smg) and borders the superior temporal gyrus at a region often referred to as the temporoparietal junction (TPJ). B. The lunate and intraparietal sulci are opened up to show the locations of several extrastriate areas in addition to the visually responsive areas within the intraparietal sulcus. These include the parieto-occipital area (PO), the medial intraparietal area (MIP), the lateral intraparietal area (LIP), the ventral intraparietal area (VIP), and the anterior intraparietal area (AIP). (Adapted from Husain and Nachev, 2007; Colby, 1988; Bisley and Goldberg, 2010).
12
General introduction comprising the areas V6 and V6A, are involved in sensorimotor transformations. The different regions within the IPS integrate information from different senses in order to influence behavior within a specific work-frame. Region AIP, for instance, is involved in grasping movements, with neurons firing in response to object presentation and manipulation (Johnson et al., 1996). Region CIP receives strong visual input, and is thought to be involved in 3D perception of objects that can guide action (for instance through area AIP) (Sakata et al., 2003). Region VIP receives input from several sensory modalities, and is thought to represent the perception of self movement and objects in peripersonal space (Sakata et al., 1998). The most extensively researched areas within the IPS are areas LIP and MIP. Area LIP is known for its role in the planning of eye movements, in particular rapid eye movements, called saccades. MIP, together with POS constitutes the parietal reach region (PRR), which is thought to provide the motor system with goal representations for reaches. All these differences have taken to suggest that PPC construct no unitary representation of space, but rather code a multitude of space representations in an effector-specific fash-
ion. In the next sections, I will discuss both LIP and MIP in more detail. Monkey area LIP and eye movements The lateral intraparietal area lies on the lateral bank of the IPS. It receives its main inputs from the occipital cortex, but also from frontal areas like the frontal eye fields. In return, LIP also projects to the frontal cortical oculomotor areas, such as the frontal and supplementary eye fields and the dorsolateral prefrontal cortex. Like many of the visual areas in the occipital cortex, studies have shown that area LIP, is organized in a topographic fashion (Blatt et al., 1990; Ben Hamed et al., 2001; Patel et al., 2010), meaning that every neuron has a preferred patch of visual space, with its neighbours covering the adjacent pieces of space. Neuronal responses – the number of action potentials that are generated – are strongest when a visual stimulus is presented in the middle of this patch of visual space, and will weaken further away from this center, up until it remains silent. This spatial field in which the neuron is most responsive is also called the neuron’s response field or receptive field (RF). In most LIP neurons, the RF is fixed to the direction of gaze, and it usually lies contralateral to the
Figure 1.3 Spiking activity of an intended movement cell during saccades to the remembered location of a visual target, presented within the cell’s response field. Each line includes responses for 8--10 trials. Trials are grouped according to increasing response delay times (denoted by increasingly darker shades of blue). The red column indicates the time of stimulus presentation. The vertical lines indicate the time at which the fixation spot was extinguished. Note that, independent of delay length, stimulus presentation evokes a strong response that is followed up by a weaker but sustained response that does not fade until the eye movement is made. (Adapted from Gnadt and Andersen, 1988)
13
1
Chapter 1
1
measured hemisphere. LIP’s functional role is still heavily debated. For at least three decades, interpretations of LIP data have been biased in two directions: one supporting a role for LIP in visuo-spatial attention (see Colby and Goldberg, 1999, for a review), and one proposing a role for the PPC in movement intentions (see Andersen and Buneo, 2002, for a review). Given its relative position between visual and motor areas, LIP is anatomically at the heart of the transformation of sensory spatial information to motor commands. Indeed, area LIP has been associated with both visual responses as well as motor related activity (Andersen et al., 1987). In addition, strong electrical stimulation of LIP neurons also elicits eye-movements into their response fields (Thier and Andersen, 1998), and vice versa reversible lesions of LIP disrupt saccade execution into the
Figure 1.4 A persistent neuron with memory activity in both visual and motor memory antisaccades. Red column represents the time of stimulus presentation. The vertical line at 1.0 s is the time of the go-cue. The colors of the lines correspond to the borders of the schematic representations of the four trial types on the top. Dark blue, prosaccade into the neuron’s response field. Light blue, an antisaccade with the stimulus presented into the neuron’s response field, and thus with the saccade away from the response field. Note that visual activity is initially the same as in the prosaccade condition, but drops during the memory delay. Red, antisaccade trials with the eye movement into the neuron’s response field. Magenta, prosaccade away from the neuron’s response field. (Adapted from Zhang and Barash, 2004)
14
neuronal group’s response field (Li and Andersen, 1999). In contrast, Goldberg and colleagues have stressed the visual and attentive responses of PPC and LIP neurons (Bushnell et al., 1981; Colby et al., 1996), arguing that spatial attention is needed before an eye movement can be made. To dissociate between the visual and motor hypotheses, LIP neurons were tested when monkeys were trained on a memory-guided saccade task, a task in which visual stimulation is separated from the saccade by some delay period in which the monkey maintains central fixation (Gnadt and Andersen, 1988, see Figure 1.3). During this delay, the LIP neurons would fire strongly in response to a saccade goal presented in their receptive field. As a marker of working memory, activity would subsequently be sustained at a lower firing rate, and then increase again around and during the time of the saccade. Gnadt and Andersen also implemented a double-step saccade paradigm, designed such that a particular neuron’s response field could be the end goal of a saccade, without ever being visually stimulated. Indeed, they found a build-up of motor related activity before and around the 2nd saccade, in the absence of visually induced activity. The authors’ interpretation of these results was that LIP encodes intentions to make a movement, rather than coding for the visual stimulus or the movement parameters. In order to disambiguate LIP’s delay activity, Zhang and Barash used the antisaccade task (Zhang and Barash, 2000, 2004). An antisaccade, as opposed to a prosaccade, is an eye movement away from a visual stimulus, with the saccade vector rotated 180 degrees relative to central fixation (Hallet, 1978). In the delayed antisaccade task by Zhang and Barash, a visual stimulus would shortly be presented within the response field of a given neuron, instructing the monkey to plan a saccade away from this response field, into another neuron’s response field. After a short delay, a go cue was given, instructing the monkey to execute the
General introduction planned saccade. They found a variety of cells; a few responded transiently to the onset of the stimulus and many to the onset of the saccade; others showed sustained activity throughout the delay reflecting either the upcoming motor direction or the visual memory (see Figure 1.4). From this, they concluded that neurons in LIP were involved in the conversion from vision to action, but that the area as a whole should be regarded neither explicitly sensory nor motor by nature (Zhang and Barash, 2004). Despite this sound conclusion, it may still be argued that during the antisaccade trials it is not the saccade vector that is being rotated, but it is actually the focus of attention that is being remapped, with attention now overlapping with the locus of the constructed saccade goal. Needless to say, it is difficult, if not impossible, to disentangle oculomotor control from spatial attention. This overlap in the two modalities provokes the question if they simply are the outcome of the same process, sharing neurons and networks along the way, differ-
ing only in actual motor output (Rizzolatti et al., 1987). In order to test this, Liu and colleagues cleverly separated saccade planning from spatial attention, by reversibly inactivating different portions of LIP while the monkeys were performing a simple saccade task or a more attentionally-demanding spatial search task (Liu et al., 2010). They found the dorsal portion of LIP (LIPd) to be more involved in saccade planning while ventral LIP (LIPv) inactivation resulted in a decrement on both saccade planning and spatial attention. However, it remains to be seen whether the discrepancy in the LIP literature can be explained by this functional-anatomical segregation of LIP. To summarize, it is well established that area LIP is tightly involved in the sensorimotor transformation; be it by directing attention to a spatial goal, by encoding the actual intention to look at the same goal, or both. At an intermediate stage between intention and attention lies the proposition that LIP may represent a spatial, gaze-centered map of salient or important stimuli in the visual
Figure 1.5 Effect of an impending saccade on visual responsiveness. Diagrams for each condition show the fixation point (cross), visual stimulus (circle), receptive field (gray area), and saccade (arrow). Time lines below show the horizontal eye position (eye) and beginning and end of stimulus (stim). The neuronal responses are averaged spikes for 16 consecutive trials. Rasters and lines are aligned on the event indicated by the long vertical line. A. Visual response to stimulus in receptive field in fixation task. B. Visual response to a stimulus that is initially outside of the receptive field. The visual stimulus and the new fixation target appear simultaneously. Left trace aligned on stimulus appearance, right trace on beginning of saccade. Note that the discharge precedes the saccade. C. Response after a saccade that removes stimulus from receptive field. Raster aligned on beginning of saccade. Note the truncation of response relative to the response to the disappearance of the stimulus in A. (Adapted from Duhamel et al., 1992)
15
1
Chapter 1
1
surrounding that can be read out by the oculomotor system, but also by regions involved in spatial attention (for reviews see Bisley and Goldberg, 2010; Fecteau and Munoz, 2006). A consequence of the gaze-centered nature of this priority map is that it needs to be updated every time the eyes move; something that has been shown in the context of a saccade task (Mazzoni et al., 1996), as well as a covert attention task (Duhamel et al., 1992). More specifically, Duhamel and colleagues showed that LIP neurons start to fire in response to their postsaccadic response field just prior to the intended intervening saccade (see Figure 1.5). This demonstrates that LIP’s spatial map is proactively updated; making it spatially stable and a reliable source for goal directed actions and spatial attention. Besides a spatial function, non-spatial functions have also been identified in the firing patterns of LIP neurons. Among many, decision making (Gold and Shadlen, 2003), reward processing (Platt and Glimcher, 1999), the passage of time (Janssen and Shadlen, 2005; Leon and Shadlen, 2003), direction of stimulus motion (Churchland et al., 2008), probabilistic reasoning (Yang and Shadlen, 2007), task rules (Stoet and Snyder, 2004) and effector use in absence of a spatial stimulus (Dickinson et al., 2003) are reflected in LIP activity. It remains an open question whether LIP activity actually encodes these parameters, or if it reflects a collateral effect on processing of space in terms of attention, priority or intention. Alternatively, nonspatial signals in LIP may reflect feedback signals from computations performed elsewhere (Balan and Gottlieb, 2009). Monkey area PRR and reaching From a sensorimotor point of view, a reaching movement to a spatial target is not so different from a saccadic eye movement: from a rich visual environment a target needs to be selected and transformed into a motor command. However, before one can reach to a target, several factors need 16
to be taken into account. First, visual spatial goals enter the brain in a gaze-centered reference frame. Since the eye can rotate in the head, and the head can in turn be rotated relative to the trunk, there is no one-to-one correspondence between this map and the innate coordinate frame of arm movements. Second, the current position of the arm is highly variable. If a certain object needs to be reached a leftward or a rightward arm movement may be needed, perhaps depending on which arm is closest to the target. In short, target location information needs to be combined with the position of the eyes, head, trunk and the arm before the correct movement can be made (see Figure 1.1). One of the first monkey electrophysiology studies showed a role for the PPC in reaching movements (Mountcastle et al., 1975). Later, this was narrowed down to area 5 and the parietal reach region (PRR), which roughly consists of areas MIP and V6A, and some parts of area 7 and PE (Buneo and Andersen, 2006; Galletti et al., 2003). An exclusive role of PRR in reaching movements was demonstrated by Snyder and colleagues, who recorded PPC neurons while monkeys performed a memory guided reach task or a memory guided saccade task (see Figure 1.6) (Snyder et al., 1997). Snyder and colleagues hypothesized that if PPC neurons mediate spatial attention, there should be no difference between delayed saccade trials and delayed reach trials, given that they require the same amount of spatial attention. Conversely, should PPC really encode intentions, they expected to find strong delay responses in the saccade trials but not in the reach trials. What they found was a mixture of neurons, some of which responded in anticipation of a saccade, while some neurons showed reach-related activity. The fact that they found a degree of anatomical separation between the two classes of neurons led them to conclude that there exist distinctly different neuronal modules for different action types within the PPC: LIP for eye movements, and PRR for reaching
General introduction movements. This dissociation has been confirmed in the absence of a spatial stimulus (Calton et al., 2002) and also when the monkey is free to choose between a saccade and a reaching movement to a spatial target (Cui et al., 2007). Based on these and other findings (reviewed in Buneo and Andersen, 2002) PRR has been suggested to encode intentions for reaches. But in what reference frame? Obviously, the most stable way to represent reach intentions would be in a bodycentered scheme, that is, in a way that information is in the native coordinate frame of the limbs. Surprisingly this is not the case. PRR’s spatial organization is best described in gaze-centered coordinates (Batista et al., 1999), meaning that targets are, just as in LIP, encoded relative to the direction of gaze. This implies that it is not the actual armmovement metric that is being encoded, but the intention to move the arm to a spatial location at a high degree of abstraction. In support of this claim, recent studies have shown that PRR activity reflects
Figure 1.6 Responses of two effector selective neurons during a delayed saccade (blue) and delayed reach (red) task. A. LIP neuron. Red bar represents the stimulus presentation time (onset at t=0s). The second vertical line (at t=0.9 s) represents the time of the go cue. B. PRR neuron. Same convention as in A. (Adapted from Snyder et al., 1997)
target selection (Scherberger et al., 2007), remaps reach goals during anti-reaches (Gail et al., 2006) and can represent multiple targets when they are part of a sequence of reaches (Balldauf et al., 2008). However, PRR is not indifferent to which arm is being used (left or right arm): neurons in PRR fire more vigorous in response to a spatial stimulus when there is an intention to move with the contralateral arm (Chang et al., 2008), placing monkey PRR at a later hierarchical level of the sensorimotor transformation than previously thought, without encoding the actual movement. But remember that the gaze-dependent reachgoal representations still need to be integrated with the position of the eyes, head, trunk and especially the arm before the correct movement can be made. If PRR activity reflects movement intentions in a gaze-centered reference frame, how then is the gaze-centered target representation translated into a body-centered muscle-based signal for motor execution (Pesaran et al., 2006; Wise et al., 1997)? This is still debated. A role for area 5 of the PPC has been suggested. In contrast to PRR, area 5 has body-centered cells as well as gaze-centered cells, meaning that this area encodes reach goals relative to our line of sight, as well as relative to the location of the hand (Buneo et al., 2002). In fact, when mapping the cortex from area 5 to PRR, Buneo and colleagues found a gradual change from body-centered to gaze-centered cells, implying some form of a topographical functional organization (Buneo and Andersen, 2006). But the question remains unanswered how and where the brain arrives at body-centered representations from gaze-centered inputs. To answer this question, one hypothesis has regained popularity in recent years. Already in the eighties it was reported that parietal neurons modulate their activity as a function of eye position (Andersen et al., 1985), but also head-, and handposition modulations have been reported (Brotchie et al., 1995; Chang et al., 2009). These modulations express themselves as a gain-change; that is, an 17
1
Chapter 1
1
increase or decrease of the firing rate of individual neurons, without distorting the spatial tuning of the recorded neurons. As such, gain-fields have a weighting effect, controlling the influence of individual neurons on the population output (Blohm and Crawford, 2009). So, even though PRR neurons primarily encode targets relative to the direction of gaze, modulations in this firing rate implicitly contain information on eye-, head- and body-posture, providing a mechanism to switch between the different reference frames and bias output to areas such as area 5, but also (pre)motor areas. Since PRR encodes predominantly reach goals with respect to gaze, but also tracks hand position in its gain fields, it has been hypothesized that area 5’s body-centered representations could come from a vectorial subtraction of the target and effector in a gaze-centered reference frame. However, skeptics have pointed out that the mechanism of gain-fields requires proprioceptive feedback signals, as well as spike integration, both of which are likely too slow for integrating body postures with spatial goals for quick and accurate movements in an ever changing environment (Wang et al., 2007). Human posterior parietal cortex and sensorimotor transformations Most of what we know about the role of the human PPC in sensorimotor processing stems from patient studies and functional neuroimaging. When confronted with severe damage to the PPC in one or both hemispheres, patients show an attentional neglect of the contralateral visual field (hemineglect), a concentric shrinking of space in visual attention, resulting in the inability to perceive more than one object at the same time (simultanagnosia), and/or a deficit in visually guided hand movements (optic ataxia) (see Pisella et al., 2009, for a review). Because optic ataxia can exist without the other two, a first conclusion can be drawn that in humans spatial attention and sensorimotor trans18
formations rely on different neuronal modules within the PPC. With the advent of fMRI, the physiology of the healthy human brain could be investigated. With fMRI, local blood oxygenation is measured, which is considered a correlate of neuronal activity (the Blood Oxygenation Level Dependent signal, or BOLD signal). Its spatial resolution is quite good, usually in the order of ~3 mm3, with every volumetric pixel (or voxel) containing millions of neurons. This, of course, is nowhere near the spatial resolution of single cell recordings. On the other hand, with fMRI it is possible to measure the entire brain at once, while with single cell recordings only a small sample of the total population of neurons is
Figure 1.7 Superior parietal cortical area with a map of remembered location (dotted circles). Subjects performed a delayed saccade task, while the polar angle of the remembered target was gradually changed. The folded and unfolded right hemisphere of a single person is shown in a posterior view. The main sulci (dark gray) have text labels. IPS, intraparietal sulcus; STS, superior temporal sulcus; POS, parieto-occipital sulcus. The small region indicated by the dotted circles is just beyond the extreme medial tip of the intraparietal sulcus. It has a strong periodic response to the task and a clear map of contralateral remembered targets (red, upper left visual field; blue, left horizontal; green, lower left). (Adapted from Sereno et al., 2001)
General introduction
1
Figure 1.8 A bilateral parietal region shown on an inflated representation of the brain, mediates gazecentered spatial updating in a double-step saccade task. Two stimuli (stim), flashed either on the left or on the right hemifield, cause increased activity in the contralateral parietal area. After a 7 sec delay, the subject makes the first saccade (sac1) and another 12 sec later the second saccade (sac2). After the first saccade, the remembered target of the second saccade switches hemifields (left-to-right or right-to-left). Correspondingly, the region’s activation also shifted: if it shifted into the contralateral hemifield, a high sustained activation was observed prior to the second saccade, but if it shifted to the ipsilateral hemifield the post-saccadic activity level decreased. (Adapted from Medendorp et al., 2003, 2008)
measured. FMRI’s disadvantage is its poor temporal resolution, making it difficult to address temporal development of brain activity. Like monkey area LIP, a portion of the human IPS was found to be involved in eye-movement planning (Muri et al., 1996), with space encoded in a retinotopic map (Sereno et al., 2001), (see Figure 1.7). Sereno’s interpretation of a retinotopic map implies that all spatial locations are encoded relative to gaze. However, without varying gaze-position, alternative explanations like a head-centered coding scheme cannot be excluded. Simply put, to maintain a correct representation of space, a gazecentered organization predicts that the activity in the human PPC needs to be updated every time the eyes move. This has been confirmed using spatial updating tasks (see Figure 1.8) (Medendorp et al., 2003; Merriam et al., 2003; Morris et al., 2007). In further analogy to monkey LIP, the human IPS
is also active during antisaccades (Connolly et al., 2000), encoding both the visual stimulus as well as the end-goal of the saccade, with a stronger activation for contraversive saccades (see Figure 1.9) (Medendorp et al., 2005; Curtis and Connolly, 2008). Similar analogies between monkey and human PPC have been reported in the context of reaching movements, with the PPC being involved in the encoding of contralateral reach goals (Connolly et al., 2003; Astafiev et al., 2003), and the integration of these goals with effector information (Medendorp et al., 2006; Beurze et al., 2007). Furthermore, the gaze-centered topographic organization of human PPC for reaches was revealed by the retuning of activity of reach goals after an intervening eyemovement (Medendorp et al., 2003). Furthermore, reminiscent of the monkey’s gain fields, eye position was found to modulate PPC activity related 19
Chapter 1
1
to pending pointing movements (DeSouza et al., 2000; Balslev and Miall, 2008). In contrast to the discrete subdivision between reach and saccade regions in monkey parietal cortex, human PPC shows, if any, a more subtle and gradual functional organization (Simon et al., 2002; Astafiev et al., 2003; Levy et al., 2007; Hagler et al., 2007; Tosoni et al., 2008; Beurze et al., 2009). Nevertheless, subdivisions of the PPC have been found based on effector biases. Furthermore, in recent years, as much as up to 6 separate topographical regions have been identified along the IPS, referred to as IPS0 to IPS6, with IPS0 also sometimes referred to as V7 (see Figure 1.9) (for a review see Silver and Kastner, 2009). Functionally, these maps may be homologues of the different subdivisions of the macaque’s IPS. Indeed, some features are shared between maps of the different species. For example, IPS5 has a head-centered organization that responds to air-puffs to the face, reminiscent of VIP’s role in coding of self-motion and objects in peripersonal space (Sereno and Huang, 2006). Also, IPS 0 through 3 have been associated with saccade planning (Sereno et al., 2001; Schluppeck et al., 2005, 2006; Levi et al., 2007), and reach tasks (Levi et al., 2007; Hagler et al., 2007). The most posterior map (IPS0/V7, most notably) shows a small preference for saccade goals, while anteromedial maps (IPS2 and 3) show a small preference for reach goals (Levy et al., 2007; Tosoni et al., 2008). But not one area responds exclusively to one of either effector types. Rather, most maps seem to respond strongest to a traveling spotlight of spatial attention (Silver et al., 2005; Swisher et al., 2007; Silver and Kastner, 2009; Szczepanski et al., 2010). Because of this, effector specific modulations may appear only as small increments in BOLD on top of the attention-related cortical activity. This discrepancy between human and non-human primate studies may be truly neurophysiological by nature. Humans may use different neuronal modules and strategies for sensorimotor integra20
tion than non-human primates. Furthermore, there are large anatomical differences between the PPC in the two species. They differ not only substantially in size; due to a disproportionate growth of the human occipital cortex the surrounding areas like the PPC have evolved a distinctly different folding pattern (Orban et al., 2004) that may have had its effect on connectivity patterns. However, methodological issues also need to be considered. First, monkeys are over-trained on the tasks they are tested on. It is known that the parietal interface is highly plastic, and sensitive to learning, reward expectation and other cognitive factors (Clower et al., 1996; Musallam et al., 2004). Because of the long duration of training, and the importance of the liquid reward, it may be efficient for the monkey’s brain to streamline the task it is trained on in separate neuronal modules. Needless to say, this may have a profound effect on the functional-anatomical organization of the PPC. This is in stark contrast with human volunteers, who are relatively briefly trained on their tasks, if they practice at all, and have in general no task-related reward expectation. But remember that gradually overlapping functional modules are not unknown for the monkey PPC. Area 5 and the PRR undergo a
Figure 1.9 Topographic areas in human parietal cortex. In parietal cortex, area boundaries correspond to the alternating representation of either the upper or lower vertical meridian. IPS, intraparietal sulcus; SPL, superior parietal lobule. (Adapted from Silver and Kastner, 2009)
General introduction gradual shift from body-centered to gaze-centered reference frames. This gradual change is reminiscent of the human’s anterior-posterior organization, with anterior areas being involved more with effector representations, and posterior areas representing spatial goals in a gaze-centered reference frame (Filimon et al., 2010). A similar effector-representation gradient may in reality also be present in non-human primates, but compromised in the
recordings due to over-training. Second, it is difficult to directly compare singlecell studies with fMRI studies, foremost because they rely on different properties of the brain (electrical properties of neurons vs magnetic properties of oxygenated haemoglobin). Furthermore, singlecell recordings sample subsets of the total number of contributing neurons, favoring large pyramidal cells that are abundant in layer IV of the cortex,
presynaptic postsynaptic dendrite axon
dendrites
cell body
synaptic cleft
axon
Exhaust line for helium gas
Transfer siphon (liquid) for liquid helium
Outer vessel Vacuum jacket and radiation shields Inner vessel Liquid helium Sensor array
Figure 1.10 Magnetoencephalography. A. Schematical overview of a neuron. (from Schoffelen, 2007, thesis). B. The lefthand panel depicts the 275 channel CTF MEG at the Donders Center for Cognitive Neuroimaging in Nijmegen (photograph © Ivar Clemens, 2009). The right-hand panel shows a cross-section of an MEG (Parkkonen, 2010). MEG is a neuroimaging technique which measures the magnetic fields arising from electrical activity in the brain. When a number of neurons receive synaptic input to their dendrites at the same time, the postsynaptic potential (PSP) creates a magnetic field that can be measured outside the skull. However, since the magnetic fields are very small - approximately one billionth of the earth’s magnetic field - specific sensors (SQUIDS) and a magnetically shielded room are required for an MEG system.
21
1
Chapter 1
1
while fMRI assesses a correlate of compound activity of up to a million neurons per voxel over the entire brain. As a third methodological issue, the two species are typically tested under different time constraints. Single-cell recordings allow for a virtually real-time read-out of neuronal activity. In contrast, fMRI has a poor temporal resolution. Due to the sluggish BOLD response, which is delayed by about 2 seconds, and peaks at about 6 s after the event occurred, interpretations of the underlying neuronal activity are hard to make (Logothetis et al., 2000). FMRI experiments are adapted to cope with the underlying temporal hemodynamics requiring a delay of at least several seconds to be able to link BOLD responses to different aspects of the task. This contrasts strongly with electrophysiological paradigms, where a delayed movement task can rely on trials as short as 0.5 s. Still, if the assumption holds that the same sensorimotor network is equally involved in both short and long trials, this difference should not matter. However, studies with optic ataxia patients have shown a deficit on short reach trials, while long reach trials where virtually unaffected (Trillenberg et al., 2007). This suggests that different neuronal modules can be involved depending on the trial duration, making a comparison between studies with different delay times difficult. Neuronal oscillations In this thesis, I present research that probes the issues discussed above on a mechanistic level. We examined the nature and temporal dynamics of sensorimotor representations using magnetoencephalography (MEG) (see Figure 1.10). MEG is a non-invasive neuroimaging technique that measures neuronal activity with a high temporal resolution from the entire brain. MEG makes use of magnetic fields surrounding the electric currents in the brain to measure neuronal activity (for reviews see Hämäläinen et al., 1993; Hari, 1999). However, 22
not the action potentials, also known as spikes, are the main contributor to this magnetic signal. Spikes are generated when a neuron is strongly depolarized. This depolarization is brought about by a summation of incoming excitatory potentials (post synaptic potentials; PSP’s) from other neurons. When a PSP arises at a synapse on a neuron’s dendrite it generates a primary intracellular electrical current. When enough of these currents summate at the target neuron’s soma, the critical threshold is passed and an action potential is generated. It’s these intracellular currents evoked by the small PSP’s (and their secondary return currents) that contribute the most to the small magnetic field that can be picked up by the MEG’s sensors (or SQUIDs) (Hämäläinen et al., 1993). But it is only when enough dendrites of many cells are regularly arranged, and PSP’s synchronously occur at these dendrites, that a magnetic field is generated that surpasses the noise level. Furthermore, it is important to note that besides excitatory, these aligned PSP’s can also be inhibitory. It’s also possible that they do not exceed the neuron’s firing threshold, but still contribute to the measured MEG signal. The electrical return current induced by the PSPs can also be measured, non-invasively with electroencephalography (EEG, not discussed here), and invasively as the local field potential (LFP). The LFP represents extra-cellular recorded fluctuations in the membrane potential of an entire group of neurons. Because the LFP represents the mean potential from a group of neurons, the PSP-induced currents are represented the strongest in the signal when they are temporally synchronized across neurons, which is also the driving force behind the MEG signal. Thus, the LFP and the EEG/MEG signal reflect similar measures, which is helpful when comparing neuronal dynamics across species. The most prominent features in the LFP and MEG signal are the event related field (ERF, also known as the event related potential (ERP) in EEG) and oscillations. ERFs are increases in electromag-
General introduction netic activity following a certain task-driven event (for instance a visual stimulus or a hand movement). Because the ERF is relatively small and typically does not outgrow the noise-level on a single trial, it is custom to average activity over a number trials. Being time-locked to the onset of a certain event, this averaging procedure filters out the noise while enhancing the stimulus evoked components of the EEG or MEG signal. Neuronal oscillatory activity can be defined as rhythmic synchronous activity of a group of neurons. These rhythms are in general not phaselocked to an experimental event, causing them to be filtered out in ERF-type averaging over trials. Instead, oscillatory activity is assessed by taking the single-trial power spectra and averaging these over trials, enhancing the induced, rather than the evoked components of the EEG or MEG signal while suppressing noise in the frequency spectrum. Oscillatory activity has been observed throughout the brain in different frequency bands (see Table 1) and are thought to provide the brain with an efficient mechanism for general processes, such as inter- and intra-areal communication, working memory, memory consolidation, functional inhibition, attention, and so forth (Buzsaki, 2006). Below, I will give a description of the two frequency bands, the gamma and alpha band, which are most relevant in light of this thesis. The gamma band In any network of inhibitory and excitatory neurons, oscillatory activity in the gamma band range can emerge (Fries et al., 2009). This rhythm is mainly orchestrated by inhibitory neurons, which impose rhythmic shunting inhibition on each other, but also on the excitatory pyramidal cells (Bartos et al., 2007), leaving only small time windows open for the pyramidal cells to be excited. Importantly, all local excitatory neurons are enslaved to the same rhythm, increasing the probability of synchronized firing of these neurons (Gray et al.,
1989). Because postsynaptic integration times are short, these localized changes in synchronization may serve to amplify behaviorally relevant signals (perceived as attention) to the cortex (Fries et al., 2001). Also, this synchronized activity could lead to non-linear increases in input gain in downstream neurons, as such facilitating communication (Salinas and Sejnowski, 2001). Furthermore, if the receiving neuron is oscillating in the same frequency as the source neurons, inputs could arrive at the right time in the oscillatory cycle, increasing the gain of the information transfer even further (Fries et al., 2005). This mechanism has also been shown to form the basis for attention, establishing synchrony between higher order areas and sensory/associative areas encoding the attended feature (Gregoriou et al., 2009). Besides communication, neuronal oscillations in the gamma range have also been implicated in active maintenance of representations in working memory (Jensen et al., 2007). When information needs to be kept in working memory, the brain relies on persistent neuronal activity (Fuster et al., 1971). How neurons sustain their activity is not fully understood, but it has been proposed that recurrent activity could be the foundation of sustained working memory representations (Hebb, 1960), with oscillatory activity as the mechanism behind this notion (Jensen et al., 2007). Indeed, human EEG and MEG studies have demonstrated a role for the gamma band in working memory maintenance (Tallon-Baudry et al., 1998; Howard et al., 2003; Kaiser et al., 2003; Jokisch and Jensen, 2007). Also in the context of sensorimotor integration, the importance of gamma-band oscillatory activity has been demonstrated. In 2002, Pesaran and colleagues showed that neurons in LIP synchronously fire in phase with a local gamma band during the memory period of a delayed saccade task. Importantly, neurons only engaged in coordinated firing when the stimulus was placed in their receptive 23
1
Chapter 1
1
Figure 1.11 Spectral signatures of delayed saccades and delayed reaches in LIP and PRR respectively. A. Spectral estimate of the local field potential measured in area LIP during a delayed saccade task. Vertical lines, stimulus presentation and go cue. Left panel, planning an eye movement into the response field of a local assembly of neurons (preferred) is accompanied by a sustained synchronization of neuronal activity in a 50-100 Hz gamma band. The right panel demonstrates the spatial selectivity of this effect, during eye movement planning to a location outside the response field (non-preferred). (adapted from Pesaran et al., 2002) B. Spectral estimate of the local field potential measured in area PRR during a delayed reach task. Vertical lines, stimulus onset, offset and go cue. Planning a reach into the response field of the local neural assembly is subserved by sustained synchronized neuronal activity in a 20-40 Hz frequency band, as demonstrated in the left-hand panel. Tuning selectivity is demonstrated by the right hand panel, where planning of a reaching movement outside the response field is accompanied by less synchronization. (Adapted from Scherberger et al., 2005)
field, showing that neurons synchronize their activity to form spatially selective memory fields (see Figure 1.11A) (Pesaran et al., 2002). In analogy, PRR neurons were shown to synchronize their activity in relation to a reach goal (Scherberger et al., 2005, see Figure 1.11B). Because these studies imposed fixed delays between vision and action, 24
no action preparation was required for the monkeys, meaning that the observed sustained gamma power increases were most likely due to the working memory for the movement goal. These findings were recently extended to human saccade preparation in a double-saccade task using MEG, demonstrating contralateral gamma band synchronization over parietal areas after stimulus presentation and just prior to the actual saccades (Medendorp et al., 2007). Oscillatory activity has also been shown to serve inter-areal communication in sensorimotor integration (Pesaran et al., 2008). Striking in this study is the timing of communication. Monkeys were trained on a sequential reach task, while neuronal and LFP responses were measured in PRR and the premotor cortex. They found spiking activity of one area to phase-lock with the LFP of the other area only just after visual stimulation and around the onset of the movement; in other words, when visual information entered the brain or a motor command was sent out of the brain the areas communicated. This complements earlier mentioned studies, where spikes locked to the phase of a local rhythm throughout the entire retention period to encode a working memory representation. Although not conclusive, this gives some support to the notion that working memory and neuronal communication are assisted by a common mechanism, namely neuronal oscillations. The alpha band The most dominating frequency in the EEG/MEG signal is the alpha rhythm (8-12 Hz). It is so dominant, because it is produced by a large patch of cortex, predominantly from occipital and parietal areas (Hari and Salmelin, 1997; Pfurtscheller et al., 1996). Being most pronounced when the eyes are closed, the alpha band was first considered an idling rhythm of the occipital cortex. This idling theory has led to the notion that the alpha rhythm is influenced by thalamocortical neurons, which in
General introduction turn are involved in sensory gating. In addition, recent work has demonstrated the importance of intracortical mechanisms in alpha oscillation generation (reviewed in Palva and Palva, 2007). Recent studies have also shown more subtle cognitive manipulations of the alpha band. For example, alpha band power increases with memory load over occipito-parietal areas (Jensen et al., 2002; Tuladhar et al., 2008), and alpha band power also increases in the hemisphere ipsilateral to an attended visual hemifield (Worden et al., 2000; Thut et al., 2006; Wyart and Tallon-Baudry et al., 2008). These and other observations have led to the hypothesis that the alpha band might be involved in functional inhibition, meaning an active disengagement of cortical areas not required in a particular task (Klimesch et al., 2007; Cooper et al., 2003; Meeuwissen et al., 2010; but see Palva and Palva, 2007). This proposal predicts for instance that successful visual perception is dependent on alpha power at the time of visual stimulation, with visual perception being most successful when alpha levels are low. Indeed, this has been confirmed in a visual detection task using MEG (Van Dijk et al., 2008). Although alpha band power is dominant over posterior areas, it is not exclusive to the visual stream. Inhibitory alpha effects have also been observed in the auditory cortex (Krause et al., 1996; Van Dijk et al., 2010) and the somatosensory cortex (Haegens et al., 2010). In the sensorimotor domain, there are two studies that have assessed the involvement of the alpha band. Okada and Salenius (1998) found a lateralization of alpha, reminiscent of afore mentioned spatial attention effects, with a relative increase of alpha over ipsilateral occipito-parietal areas relative to the direction of an upcoming eye movement. Medendorp and colleagues extended these findings with a double-saccade paradigm, where subjects had to remember one or two sequentially presented spatial stimuli (Medendorp et al., 2008). During the delay following the presentation of the
first spatial stimulus, they found a similar alpha lateralization as Okada and Salenius, with a stronger desynchronization over contralateral occipital and parietal areas. After presentation of the second spatial stimulus, at which stage two spatial stimuli had to be kept in working memory, they found a further depression of alpha band power in the same areas, effectively showing a role for alpha oscillations in spatial working memory maintenance and saccade planning. Medendorp and colleagues concluded that the alpha band is involved in allocating resources to task-relevant regions during sensorimotor transformations by inhibiting task-irrelevant areas. This Thesis Although it is well established that the human PPC plays an important role in saccade and reach planning (Sereno et al., 2001; Medendorp et al., 2003, 2005), little is known about the underlying neural mechanisms and temporal dynamics of PPC activity. As discussed above, evidence for a role of gamma band synchronization has been established in monkey area LIP (Pesaran et al., 2002) and PRR (Scherberger et al., 2005). However, whether this is also true for human PPC is still an open question. In this thesis we present a series of MEG experiments to shed light on the temporal dynamics and the underlying neural mechanisms of the healthy human PPC during sensorimotor integration. First, we studied oscillatory activity during the planning of eye movements, taking the results of Pesaran et al. (2002) as the starting point. Their results, however, did not reveal the nature of the memory representation coded in the spectral activity, whether it reflects a retrospective code of the visual working memory, or a prospective code, encoding the motor vector for the upcoming saccade. In the 2nd chapter of this thesis we address the following questions: Do human PPC neurons engage in rhythmic activity to encode saccade goals, and if so, does this memory trace represent the work25
1
Chapter 1
1
ing memory of the stimulus, or the encoding of an upcoming eye-movement? We address these questions by recording MEG signals from human volunteers while they are performing delayed proand antisaccades. These issues, and the observed results, are put into a broader perspective in the 3rd chapter, when we review literature on prosaccades and antisaccades from a micro- to a macroscopic view, building a case for oscillatory activity as a common ground on which the different species and measurement techniques can be compared. In the 4th chapter we tested the reference frames in which spectrally-tuned activity for saccade planning is encoded. Using a spatial updating task, we addressed the question whether these spatial representations are coded in gaze-centered coordinates, requiring updating after every eye movement, or whether they are stored in a gaze independent reference frame, in which case intervening eye movements should be of no concern. To answer these questions, we have used a classic double-saccade paradigm, in which stimulus and saccade are temporally separated by two delay periods and an intervening eye movement. By balancing our design on stimulus position, gaze position
26
and final saccade amplitude, we were able to dissociate the contributions of the different components to the power spectrum. In the 5th chapter, we address the issue of effector specificity of the different parietal regions. As discussed above, the monkey IPS seems clearly subdivided in effector-related regions, while this subdivision is less clear-cut in the human PPC. In this chapter, we take a different perspective on this problem. We hypothesize that spatially intermingled populations of neurons in the IPS might be grouped together by effector-dependent oscillatory activity. First, we address this hypothesis by asking if reach goals are represented in oscillatory activity by human PPC neurons. Next, we ask the question whether reach goals are represented by spectral power in a different frequency band than saccade goals. As a last issue in this chapter, we have localized the different sources of the different effector specific frequency bands, testing whether they originate from the same or different modules in the PPC. Finally, in the 6th chapter, we summarize our findings and briefly discuss the implications they may have for future research.
General introduction
1
27
28
Chapter 2 Gamma-band activity in human posterior parietal cortex encodes the motor goal during delayed prosaccades and antisaccades
Adapted from Van Der Werf j, Jensen O, Fries P, Medendorp WP, (2008) Gamma-band activity in human posterior parietal cortex encodes the motor goal during delayed prosaccades and antisaccades. Journal of Neuroscience 28:8397-8405. 29
Chapter 2
T
2
he parietal cortex has been implicated in coding spatial memories for saccades. Human fMRI and monkey electrophysiological studies have recently revealed that these memories are coded in topographic maps storing locations of targets in the contralateral hemifield (human: Sereno et al., 2001;Schluppeck et al., 2005;Jack et al., 2007; Medendorp et al., 2003,2006; monkey: Blatt et al., 1990;Ben Hamed et al., 2001;but see Platt and Glimcher, 1997). From a mechanistic perspective, suggestions have been made that rhythmic activity provides the neural basis for the maintenance of spatial memory representations (Ward et al., 2003; Tallon-Boudry et al., 1998; Pesaran et al., 2002; Jensen et al., 2007). Rhythmic neuronal synchronization can be measured invasively as local field potentials (LFP), but also as electric potentials or magnetic fields on the scalp (Varela et al., 2001). Based on these measurements, it has been conjectured that oscillations in the gamma band (>30 Hz) mediate the active maintenance of a working memory (Tallon-Baudry et al., 1998; Jensen et al., 2007) whereas activity in the alpha band (7-13 Hz) reflects a regulatory mechanism disengaging areas not relevant for a given process (Klimesch et al., 2007; Medendorp et al., 2007; Jensen et al., 2002; Jokisch and Jensen, 2007; Sauseng et al., 2005; but see Palva and Palva, 2007). Few studies have characterized oscillatory brain activity during the memory period in delayed-saccade tasks (Okada and Salenius, 1998; Pesaran et al., 2002; Medendorp et al., 2007; Lachaux et al., 2006). Recently, Medendorp et al. (2007) reported parieto-occipital power suppression of the alpha band in the hemisphere contralateral to the stimulus. They also observed contralateral power enhancements in the gamma band (60-100 Hz), analogous to the spatially-tuned gamma band power in LFP activity observed during working memory in monkey parietal cortex (Pesaran et al., 2002). It is unclear, however, whether these parietal oscilla30
tions relate mainly to the processing of spatial sensory information or to the coding of the motor goal of the saccade. In this study, we used a memory-guided antisaccade task to address this question. The antisaccade task dissociates sensory from motor goal representations as it requires participants to transform the location of a stimulus into an eye movement to the opposite visual field (Hallett, 1978; Munoz and Everling, 2004). A number of studies have reported that the parietal cortex is engaged in spatial remapping for antisaccades (Zhang and Barash, 2000,2004; Everling et al., 1998; Moon et al., 2007; Medendorp et al., 2005), but the temporal structure of human parietal activity in relation to the neural dynamics of this process has not been revealed. We applied magnetoencephalography (MEG) to record oscillatory brain activity from human subjects instructed to plan either pro- or antisaccades. By exploiting the hemispheric lateralization of the power in the various frequency bands, we analyzed the time-varying aspects of sensorimotor processing during saccade planning. Our results show a dissociation between alpha- and gamma-power in parietal areas. While an alpha-power decrease was primarily stimulus-related, a gamma-power increase was only initially stimulus-related, but subsequently corresponded to the saccade goal.
Methods Participants 19 healthy subjects (3 female, 16 male; mean age 26 +/- 3 years), free of any known sensory, perceptual, or motor disorders, volunteered to participate in the experiment. All subjects provided written informed consent according to institutional guidelines of the local ethics committee (CMO Committee on Research Involving Human Subjects, region Arnhem-Nijmegen, the Netherlands).
Gamma-Band Activity during Antisaccades
MEG Recordings Subjects were seated upright in the MEG system that was placed in a magnetically shielded room. They were instructed to sit comfortably without moving, and to look at the stimulus screen, located about 40 cm in front of them. Visual stimuli, generated with Presentation 9.10 software (Neurobehavioral Systems Inc., Albany), were presented using a LCD video projector (SANYO PROxtraX mutiverse, 60 Hz refresh rate) and back-projected onto the screen using two front-silvered mirrors. MEG data were recorded continuously using a wholehead system with 151 axial gradiometers (Omega
2000, CTF Systems Inc., Port Coquitlam, Canada). Head position with respect to the sensor array was measured using localization coils fixed at anatomical landmarks (the nasion and at the left and right ear canal). These measurements were made before and after the MEG recordings to assess head movements during the experiment. In addition, horizontal and vertical electrooculograms (EOG) were recorded using electrodes placed below and above the left eye and at the bilateral outer canthi. Electrode impedance was kept below 5 kΩ. During the experiment, these recordings were continuously inspected to ensure the subject was vigilant and per-
Figure 2.1 Experimental paradigm. A. Sequence of stimuli and subject instructions. Each trial started with a baseline period of 1.5 s, during which subject was instructed to fixate a central cross. Next, a peripheral stimulus (S) was flashed (for 0.1 s), followed by a 1.5 s delay interval. Subsequently, the fixation cross was turned off, signaling the subjects to look either toward the remembered location of the stimulus on pro-saccade trials or to its mirror location in the opposite hemifield on anti-saccade trials and then back to center when the fixation cross reappeared. Corresponding eye position traces (horizontal component) of one typical subject are shown in green and red, respectively. B. Outline of the entire experiment. Subjects performed blocks of 20 consecutive pro- or anti-saccade trials, between which a brief rest (10 s) was provided. The type of saccade (pro/anti) to be made was instructed prior to the start of each block of trials.
31
2
Chapter 2
2
formed the task correctly. MEG and EOG signals were low-pass filtered at 300 Hz, sampled at 1200 Hz, and then saved to disk. Prior to the actual measurements, the EOG signal was calibrated using a 9-point calibration grid. For each subject, a full brain anatomical MR image was acquired using a standard inversion prepared 3D T1-weighted scan sequence (FA = 15°; voxel size: 1.0 mm in-plane, 256 x 256, 164 slices, TR = 0.76 s; TE = 5.3 ms). The anatomical MRIs were recorded using a 1.5 T whole-body scanner (Siemens, Erlangen, Germany), with anatomical reference markers at the same locations as the head position coils during the MEG recordings (see above). The reference markers allow alignment of the MEG and MRI coordinate systems, such that the MEG data can be related to the anatomical structures within the brain. Experimental Paradigm Subjects made delayed pro- and antisaccades using a block design paradigm. Pro- and antisaccade blocks were alternated with 10 s blocks of rest in between, during which subjects could freely move their eyes (Figure 2.1). The type of upcoming saccade block (Pro or Anti) was indicated by a letter (P, or A) in the center of the screen at the end of each resting block. Each trial within a pro- or antisaccade block started with a subject fixating centrally on a white cross (see Figure 2.1A). Then, after a baseline period of 1.5 s, a peripheral white stimulus dot was flashed for 0.1 s, at a random eccentricity between 9° and 18° and at a random angular elevation within a range of -36º and 36º in either the lower or upper visual field. This was followed by a 1.5 s memory delay during which the subject maintained fixation. Subsequently, the fixation cross was turned off, instructing the subject either to saccade toward the remembered location of the stimulus in prosaccade trials, or to its mirror location (relative to the fixation point) in the opposite hemifield during anti32
saccade trials. Then, 0.3 s later, the central fixation cross was turned on again, instructing the subject to fixate at the centre of the screen till the end of the trial. We purposely used a short time interval between the task-related saccade and the return saccade in order to maximize the number of trials performed during the experiment, and hence to achieve maximal statistical power in the study. The paradigm had four different conditions: a prosaccade to a right hemifield stimulus (PR), an antisaccade on a right hemifield stimulus (AR), a prosaccade to a left hemifield stimulus (PL), and an antisaccade on a left hemifield stimulus (AL). Each trial lasted 4 s. Each block consisted of 20 trials, with 10 left and 10 right hemifield stimulus locations pseudo-randomly interleaved (see Figure 2.1B). In total, there were 15 blocks for each condition resulting in a final number of 600 trials, lasting about 45 minutes. Behavioral analysis Eye movements were recorded in all subjects. An example of the eye traces (horizontal component) of a typical subject during the pro-left condition (green trace) and anti-left condition (red trace) is shown in Figure 2.1A in relation to the temporal order of events. As shown, this subject maintained fixation during the baseline period, the presentation of the stimulus, and the memory interval, and made eye movements in the correct directions after the fixation spot was turned off. Eye movement recordings in all 19 subjects confirmed that they followed in most trials the instructions correctly. Trials in which subjects broke fixation, made saccades in the wrong direction or blinked the eyes during the trial, were excluded from further analysis. For the remaining trials, it was assumed that subjects had retained a veridical spatial representation during the memory interval as their saccades landed close to (or on) the ideal spot. On average 472 ± 73 (SD) trials were incorporated in the analysis of each participant. Furthermore, reaction times for
Gamma-Band Activity during Antisaccades
antisaccades (214 ± 74 ms, mean ± SD) and prosaccades (212 ± 72 ms) were not significantly different (t-test, P=0.52), which is consistent with previous results using a similar paradigm (Medendorp et al. 2005). MEG data analysis Data were analyzed using Fieldtrip software (http://www.ru.nl/fcdonders/fieldtrip), an open source Matlab toolbox for neurophysiological data analysis developed at the F.C. Donders Centre for Cognitive Neuroimaging. From the trials that survived the exclusion criteria described above, data segments that were contaminated with muscle activity or jump artifacts in the SQUIDs were excluded using semi-automatic artifact rejection routines. For the sensor level analysis, an estimate of the planar gradient was calculated for each sensor using the signals from the neighboring sensors (Bastiaansen and Knosche, 2000). The horizontal and vertical components of the planar gradients approximate the signal measured by MEG systems with planar gradiometers. The planar field gradient simplifies the interpretation of the sensor-level data since the maximal signal is located above the source (Hämäläinen et al., 1993). Power spectra were computed separately for the horizontal and vertical planar gradients of the MEG field at each sensor location and the sum of both was computed to obtain the power at each sensor location irrespective of the orientation of the gradient. Time-frequency representations (TFR), estimating the time course in power, were computed using a Fourier approach, applying a sliding tapered window. Because the gamma band is typically much wider and more variable across subjects than the alpha band (Hoogenboom et al. 2006), we analyzed two frequency ranges separately: 3-30Hz and 30120 Hz. For the lower frequency band (3-30 Hz), we applied a fixed time window of 0.5 s and a Hanning taper. This resulted in a spectral smoothing of roughly 3 Hz. For the higher frequency band (30-
120 Hz) we applied a multitaper approach (Percival and Walden, 1993) using a fixed window length of 0.4 s and 11 orthogonal Slepian tapers. This resulted in a spectral smoothing of approximately 15 Hz. We examined the task-related changes in power in various frequency bands relative to average power in the baseline periods (see Figure 2.1). The baseline power was computed over a time window (0.5 and 0.4 s for the lower and higher frequency bands, respectively) centered 0.3 s prior to the presentation of the stimulus. Using a jackknife procedure (Efron and Tibshirani, 1993), we determined the variance of the power in the selected frequency bands across trials. Using these estimates, we expressed the difference in power between the memory period and the baseline as a t-score for each subject and for each condition. The resulting t-scores were transformed into z-scores (Medendorp et al., 2007) to obtain a normalized power estimate. Using these values, the directional selectivity in the various frequency bands was examined by comparing the power in each sensor for stimuli in the contralateral and ipsilateral hemifield. The resulting z-scores, which are well normalized for intrasubject variance, were pooled across subjects (zgroup = 1/√N ∑zi with zi being the z score of the i-th subject). Groups of sensors of interest (see Figure 2.2C and 2.3C, middle panels) were selected based on their response to visual stimuli (as in Medendorp et al., 2007). Statistical significance was tested at the sensor level, using a non-parametric permutation test across subjects. Z-scores representing the contrast between the conditions were computed for each subject within the predefined channel-frequencytime window of interest, resulting in a single number per subject. Subsequently, the significance at the group level was assessed by pooling the z-scores over all subjects. Testing the probability of this pooled z-score against the standard normal distribution would correspond to a fixed effect statistic. To be able to make statistical inference correspond33
2
Chapter 2
2
ing to a random effect statistic, we tested the significance of this group-level statistic by means of a randomization procedure. We randomly multiplied each individual z-score by 1 or by -1 and summed it over subjects. Multiplying the individual z-score with +1 or -1 corresponds to permuting the original conditions in that subject. This random procedure was repeated one thousand times to obtain the randomization distribution for the group-level statistic. The proportion of values in the randomization distribution exceeding the test statistic defined the Monte Carlo significance probability, which is also called a p-value (Nichols and Holmes, 2002; Maris and Oostenveld, 2007). We defined the time interval of the sensory response from 0.1 to 0.6 s after stimulus onset. The retention period was defined from 0.6 to 1.6 s after the presentation of the stimulus, and excludes the initial sensory response. We chose our frequency ranges of interest to be 7-13 Hz for the alpha band and 70-120 Hz for the gamma band. These frequency ranges are compatible with previous reports on oscillatory activity in human saccade planning (Lachaux et al., 2006; Medendorp et al., 2007). To localize the neural sources of the different spectral components, we applied an adaptive spatial filtering or beamforming technique (Dynamic Imaging of Coherent Sources (DICS), Gross et al., 2001; Liljeström et al., 2005). Each subject’s brain volume was divided into a regular 1 cm three-dimensional grid. For each grid point, a spatial filter was constructed that passes activity from this location with unit gain, while attenuating activity from other locations (Van Veen et al., 1997). This filter was computed from forward models with respect to dipolar sources at each grid point (the leadfield matrix) and the cross spectral density between all combinations of sensors at the frequency of interest (Nolte, 2003). We used a multi-spherical volume conductor model to compute the leadfield matrix by fitting a sphere to the head surface underlying each sensor (Huang and Mosher, 1997). The head 34
shape was derived from each individual structural MRI. As for the sensor data, we computed the power changes at the selected frequency bands for each subject. Using these power estimates, we calculated the z-statistic to express the power effects across subjects at the source level. Using SPM2 (http://www.fil.ion.ucl.ac.uk/ spm), the individual anatomical MRIs and the corresponding statistical maps were spatially normalized to the International Consortium for Brain Mapping template (Montreal Neurological Institute, Montreal, Canada). The individual spatially normalized statistical maps were subsequently averaged to get an estimate of the location of the sources producing the effect in the various frequency bands.
Results We investigated the time-varying directional selectivity of power in the various frequency bands during saccade planning. Subjects were tested in four different conditions: either a delayed prosaccade or a delayed antisaccade in response to a visual stimulus presented either in the left or right visual hemifield (see Methods). We exploited the hemifield-specific lateralization of power during the memory period to discriminate representations of visual sensory information (stimulus location) and representation of motor goal information (saccade direction). Sustained parietal gamma activity contralateral to the stimulus during prosaccades We first present the results of the higher frequencies for the two prosaccade conditions, PL and PR. Figure 2.2A and B, middle panels, plot the scalp topography of the power changes in these conditions relative to baseline (see Figure 2.1), averaged across 40-120 Hz and time 0.1 – 0.6 s after stimulus offset. Regions with warmer (red) colors indicate an increase of power relative to baseline, while regions with cooler (blue) color reflect a decrease of gam-
Gamma-Band Activity during Antisaccades
2
Figure 2.2 The human PPC shows directional-selective gamma-band activity during the planning of prosaccades. A, B. Time-frequency resolved power changes relative to baseline (left and right-hand panels) of the PL and PR conditions for the sensors marked (by asterisks) in the central panels, which show the topographic distribution of power in the 40-120 Hz frequency range, averaged across subjects during the first 0.5 s after stimulus presentation. Color format: warmer (red) colors, power increase re. baseline; cooler (blue) colors, power decrease. Time t = 0 s: onset of the visual stimulus; t = 1.6 s: go cue for the saccade. C. Directional selectivity (DP) of time-frequency resolved gamma-band power during pro-saccades, for the sensors marked in the central panel. Left hemisphere: PR – PL; Right hemisphere: PL – PR. Color format: warmer (red) colors, preference for stimuli in the contralateral hemifield; cooler (blue) color, bias toward ipsilateral stimuli. D. Directional selectivity in the higher frequency bands pooled across hemispheres. Color format as in C.
ma-band power. Both topographies show a clear increase of gamma-band power in the contralateral hemisphere, most prominently over the parietal areas. Based on the strongest induced gamma and previous reports (Medendorp et al., 2007), we selected two symmetric subsets of posterior sensors (marked by asterisks) for further spectral analysis. The left- and right-hand panels of Figure 2.2A and B show the time-frequency representations of the power changes in the higher frequency band (30120 Hz) for these posterior sensors in the left and right hemisphere, respectively. Time t=0 s denotes the onset of the visual stimulus and t=1.6 s the go-
cue for the saccade. As shown, in both hemispheres, the selected sensors show strong enhancements in gamma-band power in response to contralateral stimuli, in the range of 40 to 120 Hz. These enhancements are followed by weak power changes relative to baseline during the delay periods. Next, to determine the directional selectivity of the power in the gamma band, we compared the power for prosaccades to stimuli in the contralateral and ipsilateral visual field (referred to as DP), separately for each hemisphere. In other words, for sensors in the left hemisphere, we subtracted activity for a stimulus in the left visual field from the ac35
Chapter 2
2
Figure 2.3 The contralateral-selective gamma band activity over human posterior parietal cortex encodes the direction of the saccade during delayed anti-saccades. Data in same format as Fig 2. A, B. Time-frequency representations of power in the AL and AR conditions relative to baseline. Selected sensors are marked in the central panels, which show the stimulus evoked topography during the anti-saccade trials. C. Directional selectivity (DA) of time-frequency resolved gamma-band power during anti-saccades, for the sensors marked in the central panel. Left hemisphere: AR – AL; Right hemisphere: AL – AR. D. Time-frequency representation using a pooled comparison across hemispheres.
tivity for a stimulus in the right visual field (DP=PR – PL), and vice versa for sensors overlying the right hemisphere (DP=PL – PR). Figure 2.2C, middle panel, shows the scalp topography of these power changes, averaged across 40-120 Hz and time 0.10.6 s after stimulus onset. Note that, in this color format, regions with warmer (red) colors indicate a preference for the stimuli in the contralateral hemifield; regions with cooler color (blue) show a bias toward ipsilateral stimuli. As shown, in both posterior hemispheres, there is a clear gammapower increase with respect to stimuli presented in the contralateral hemifield. Figures 2.2C, left- and 36
right-hand panels, show the time-frequency representations of the power differences (for contraminus ipsilateral stimulus locations) in the higher frequency band (30-120 Hz) for the selected posterior sensors in the left and right hemisphere, respectively. The TFRs revealed a broad-band power increase (40-120 Hz) following the stimulus. This was followed by a narrower band of sustained gamma activity (85-105 Hz) in the rest of the delay period, which was stronger over the left hemisphere. Finally, based on the symmetry of these panels, the spectrograms were pooled across hemispheres, resulting in the combined hemisphere-specific
Gamma-Band Activity during Antisaccades
changes in power for prosaccades with respect to contra versus ipsilateral stimuli (Figure 2.2D). In response to the stimulus, the parietal sensors showed a significant increase in power for contralateral stimulus locations across the 40-120 Hz frequency range. This is consistent with the scalp topography in Figure 2.2C. This selectivity was maintained within a narrower frequency band (85105 Hz) during the delay interval (0.6–1.6 s), where the gamma activity became more prevalent prior to the initiation of the saccade. A non-parametric randomization test (see Methods) over the timefrequency window from 0.6 – 1.6 s and 85 – 105 Hz (indicated in Figure 2.2) revealed the significance of this effect (p < 0.05). Sustained gamma band activity reflects stimulus-togoal mapping during antisaccades The key question here is what aspect of sensorimotor processing is encoded in the gamma band of this bilateral source during the delay period. Does the observed lateralized power during the memory period reflect the maintenance of the visual stimulus or rather the representation of the goal for the pending saccade? Obviously, the results of the prosaccade trials cannot answer this question since they do not discriminate the location of the stimulus from the goal of the saccade. However, the antisaccade trials allow us to address this question: In these trials, the coordinates of the visual stimulus must be mirrored to specify the goal of the saccade. If the observed parietal gamma-band power reflects the outcome of this process we would expect the selectivity to reverse from contralateral to ipsilateral tuning after presentation of the stimulus in order to represent the saccade goal (Medendorp et al., 2005; Zhang and Barash, 2004). However, if the gamma activity in parietal cortex encodes a visual memory representation, we would not expect a difference in the directional selectivity of power during the delays of prosaccade and antisaccade trials
in response to the same visual stimulus. Figure 2.3 presents the power in the gamma band for the antisaccade conditions, AL and AR, in the same format as Figure 2.2, using the same group of sensors. Figure 3A and B show the power differences of these conditions relative to baseline; Figure 3C plots the power differences between the conditions in terms contralateral and ipsilateral stimulus locations (referred to as DA). Focusing on the latter (DA), the TFRs of both hemispheres (left- and right-hand panels of Figure 2.3C) show a transient response after stimulus presentation which is biased toward the contralateral hemifield, as for prosaccades. The topography of the stimulusevoked gamma-power (Figure 2.3C, middle panel) is similar to that of the prosaccades (Figure 2.2C). However, in contrast to the prosaccades, this contralateral selectivity vanishes quickly and a significant bias in power (p < 0.05) toward the ipsilateral field emerges during the delay interval, at a frequency range of 85 to 105 Hz. This can be seen more clearly in the pooled spectrogram (Figure 2.3D), showing that the ipsilateral bias arises at about 500 ms after stimulus onset. Because ipsilateral visual cues define contraversive movements during antisaccades, this reversal should be interpreted as a remapping from stimulus-to-goal selectivity in the memory interval. In other words, the dynamic shift observed in the 85-105 Hz band represents a transition from a visual stimulus representation (or a default stimulus-defined response) to a motor goal representation during the course of an antisaccade trial. This suggests that directional-selectivity of gamma-band power in parietal cortex represents target locations for upcoming movements, rather than remembered locations of visual stimuli. To test whether the time of switching toward the motor goal changed during the course of the experiment, we split the antisaccade trials in a first and second half, respectively. We then compared the respective spectrograms of the first and second half, which revealed no significant difference 37
2
Chapter 2 in the time of the reversal (p>0.05). This suggests the switch for the saccade goal did not develop over time in our testing conditions.
2
Sustained gamma activity encodes the goal of the saccade Thus, in most simple terms, the results indicate contributions from two main processes to hemispheric gamma band selectivity during the planning of a saccade: an initial coding of visual sensory information (stimulus component) and a later representation of motor goal information (goal component). In order to separate the stimulus from the motor goal component, we decomposed the differential pro- and antisaccade activities (DP and DA) by assuming that they have the same stimulus (S) component (same sensory stimulation) and opposite goal (G) components (opposite saccade directions). Based on these assumptions, the pro and antisaccade selectivities equal DP = S + G and DA = S – G. Hence, the stimulus and goal components can easily be derived following respectively S = (DP + DA)/2 and G = (DP - DA)/2. Figure 2.4 presents the results of this decomposition analysis. The top panel depicts the reconstructed stimulus component (S), whereas the bottom panel shows the goal (G) component. As the
Figure 2.4 Decomposition of gamma-band activity in stimulus and saccade goal components. A. Stimulus component B. Motor goal component. The color code represents the difference in power between contralateral and ipsilateral stimulus locations (in A) / saccade goals (in B).
38
figure suggests, there is a broad-band transient stimulus-induced component (Fig 2.4A), lasting for less than 0.6 s. There was no evidence for sustained gamma activity reflecting a persisting stimulus representation (p > 0.3). The goal component (Fig 2.4B), however, showed a significant (p < 0.05) sustained component (85-105 Hz) during the entire delay interval, increasing in strength toward the execution of the saccade. Thus, there seems to be a full dissociation: the sustained gamma band selectivity encodes the goal of a saccade and not the memory representation of the previous visual stimulus. Source underlying cue-to-goal mapping in the posterior parietal cortex We used spatial filtering techniques to estimate the sources underlying the gamma band dynamics identified at the parietal sensors. Figure 2.5 present these results, using the pooled activity across hemispheres (thresholded at |z|>1), on a rendered representation of a standardized left hemisphere. The left-hand panels represent the early (0.1 – 0.6 s) stimulus-induced broad-band (40 – 120 Hz) gamma activity, while the right-hand panels show the activity during the late delay period (1.1 – 1.6 s) at the 85 – 105 Hz band. For the pro- and antisaccade conditions (Figure 2.5A and B, left panels), the early stimulus-induced broadband gamma band activity originates from occipital and posterior parietal regions, with the peak in the occipital cortex. Figure 2.5C, left panel, further validates this as stimulus-driven activity by showing that virtually the same source gives rise to the stimulus component in the decomposition analysis. By contrast, the source of the sustained directionally-selective gamma band activation was found merely in a smaller region in the posterior parietal cortex, for both pro and antisaccades (Figures 2.5A and B, right panels). Importantly, the parietal source that underlies the sustained gamma response during antisaccades is almost identical to
Gamma-Band Activity during Antisaccades
2
Figure 2.5 Source reconstructions of gamma band activity. A. Source of directionally-selective gammaband power during pro-saccades (DP). The broad-band (40-120 Hz; 0.1-0.6 s) stimulus-related source is located in occipital-parietal regions (left-hand panel). The source of the narrow-band (85-105 Hz) gamma activity during the memory period (0.6-1.6 s) is in posterior parietal cortex (right-hand panel). B. Source of directionally-selective gamma-band power during anti-saccades (DA). Parietaloccipital regions respond to the stimulus at first (left-hand panel), as for pro-saccades. During the delay-period, a region in the posterior parietal cortex shows selectivity in the 85105 Hz frequency range for ipsilateral stimulus locations, which is consistent with the coding of contraversive saccades (right-hand panel). C. Sources reconstructions of the stimulus and goal components based on the decomposition analysis. In all plots, the source activity was pooled across hemispheres, tresholded at |z|>1, and shown on a standard left hemisphere. Color format as in the corresponding TFRs.
the neural locus coding the sustained response during prosaccades (represented by blue since contra minus ipsilateral gamma-power is illustrated). This is confirmed by the locus of activity that represents the goal component (Figure 2.5C, right panel), as reconstructed by the decomposition analysis. The peak of this activity was found in the superior parietal lobe, medially from the intraparietal sulcus. This is in line with previous fMRI studies showing regions with a topographic organization for saccades at similar locations in parietal cortex (Medendorp et al., 2003; Sereno et al., 2001; Schlup-
peck et al., 2005). Contralateral alpha band suppression during proand antisaccades We also investigated the activation in the lower frequency band (3-30 Hz) using the same group of parietal sensors as used in the high-frequency analysis (marked in Figures 2.2 and 2.3). The results, expressed as a z-score, are shown in Figure 2.6, in a pooled comparison across hemispheres. The color code represents the difference in power for contralateral vs. ipsilateral stimuli. First, in the theta 39
Chapter 2
2 Figure 2.6 Time-frequency resolved power in the lower frequency band in a pooled comparison across hemispheres. A. Pro-saccades. B. Anti-saccades. Data from sensors marked in Fig 2A and 3A. Color format: blue regions indicate a stronger suppression for remembered target locations in the contralateral hemifield, and vice versa for red regions.
Figure 2.7 Decomposition of lower-frequency activity in stimulus and saccade goal components. A. Stimulus component B. Motor goal component. C. Source of the stimulus component in the alpha-band (7-13 Hz), reconstructed from the 0.5 s period after stimulus presentation, shown on a standard left hemisphere (thresholded at |z|>2). The color code represents the difference in power between contralateral and ipsilateral stimulus locations (in A) / saccade goals (in B).
40
range, at about 5 Hz, a significant contralateral transient enhancement of activity can be discerned in response to the stimulus in both saccade conditions (pro/anti, p < 0.05). Visual stimulation typically evokes responses that are time-locked to visual onset and, in spectral analysis, primarily contain power components in the theta range. The parietal sensors further show clear contralateral reduction in the alpha (7-13 Hz) and the beta bands (13-25 Hz). The beta band reduction is only transient, in both conditions, which is also most likely a marker of sensory processing. By contrast, in the alpha band, the transient responses are followed by a sustained suppression in the prosaccade condition (Figure 2.6A). For antisaccades, the contralateral suppression effect in the alpha band seems to be less sustained, but does not reverse (Figure 2.6B). In other words, in contrast to the gamma band results, the alpha band seems not so much related to the processing of the saccade goal, but is more related to coping with the sensory information. The sensory bias of the alpha-band power is further emphasized in Figure 2.7, by means of the decomposition analysis, like above. Undeniably, the alpha band is suppressed in relation to the stimulus location for the most part of the trial interval (Figure 2.7A). The analysis does show a significant alpha suppression in relation to the direction of the saccade, but only emerging in a 0.5-s period prior to the go-cue of the saccade (p < 0.05). Thus the alpha band shows a general decrease during memory-guided pro- and antisaccades, reflecting an ongoing sustained stimulus representation and a late but modest build-up of the goal representation for the saccade (Figure 2.7B). Using spatial filtering techniques, we located the sensory component in extrastriate brain regions, as a widespread pattern that extended from close to the intraparietal and parietal-occipital sulcus into anterior occipital cortex (Figure 2.7C).
Gamma-Band Activity during Antisaccades
Discussion We studied the dynamics of oscillatory activity in parietal cortex during saccade planning. The starting point of our investigation was the hemispherespecific contralateral-selectivity of spectral power in various frequency bands during memory-guided prosaccades, consistent with previous reports (Medendorp et al., 2007; Pesaran et al., 2002). The novelty of the present study lies in the use of a memory-guided antisaccade task to distinguish this spectral activity in components related to sensory processing and those reflecting a motor goal representation. Our main finding is that gamma band synchronization in posterior parietal cortex encodes the upcoming motor goal, whereas alphaband desynchronization in parieto-occipital regions is linked predominantly to the processing of the stimulus that defines that goal. Our findings suggest that the oscillations in the alpha and gamma frequency bands subserve different functions in visuomotor processing for saccades. As Figure 2.4 shows, stimulus processing in the gamma band is terminated at 0.5 s and then there is a complete goal representation of the pending saccade during the rest of the delay period, increasing in strength closer to saccade execution. The delay-period activity in the alpha band is most consistent with a long-lived stimulus-driven component superimposed on a later representation of the goal of the saccade (Figures 2.6 and 2.7). Thus, compared to the gamma band, the changes in the alpha band demonstrated a much slower time course. This argues against a strong coupling between gamma band and alpha band mechanisms in terms of their functions (Jokisch and Jensen, 2007). The quasi-static desynchronization of the alpha band during antisaccades would be consistent with a general regulatory mechanism, allocating resources for processing sensory information without actually encoding this information (Kelly et al., 2006; Thut et al., 2006; Medendorp et al., 2007; Worden et al., 2000).
The full dissociation of the gamma band selectivity in terms of representing the goal of the upcoming saccade and not the previous stimulus location is consistent with a recent neuroimaging study by Medendorp et al. (2005). Using event-related fMRI, they demonstrated a full interhemispheric shift of BOLD activity during target remapping in memory-delayed antisaccades. In this respect, our findings compare well with recent observations showing a close correlation between gamma band activity and BOLD signals in simultaneous neural and hemodynamic recordings (Logothetis et al., 2001; Niessing et al., 2005). Recently, Zhang and Barash (2000, 2004) recorded from neurons in monkey area LIP during the memory-delayed saccade task. In their paradigm, they cued stimulus and saccade type (pro/anti) simultaneously, which is slightly different from the present study in which saccade type was instructed in advance of the visual stimulus. Nevertheless, the observed effects in the gamma-band are consistent with their findings that the stimulus direction is remapped in monkey LIP in order to code the goal of an antisaccade. Furthermore, of all the neurons that remapped, Zhang and Barash reported an average inversion time of about 400 ms, even though individual neurons could remap within 50 ms. Figure 2.5 suggests an inversion time of about 500 ms, which is fairly close to their population average. Other human studies on the time course of eventrelated potentials or fields during non-delayed antisaccades reported much shorter reversal times, i.e., in the order of about 100 ms (Moon et al., 2007; Everling et al., 1998). Evidently, the delayed antisaccade paradigm as such provides no incentive for subjects to invert the stimulus location directly after its onset. This could explain the relatively long average inversion-time, although there was also clear, but weaker, goal-related activity preceding this time, as shown by Figure 2.4B. Conversely, studies that do not impose a delay between the visual stimulus and the antisaccade response may 41
2
Chapter 2
2
observe not only remapping effects but also effects of other processes involved in saccade generation (Moon et al., 2007). Importantly, Zhang and Barash (2004) observed the remapping effect in only 60% of the neurons in the recorded population. They also found neurons that coded consistently the location of the stimulus as well as neurons that exclusively encoded the movement goal. In contrast, the present human study, and the decomposition analysis in particular (see Figure 2.4), clearly shows that the goal for a saccade is represented in the parietal gamma band activity and not the memory of the visual stimulus. One potential reason for this difference is that Zhang and Barash made their claims based on neuronal firing rates. We may have probed a more distinct cognitive variable coded within this activity across a neural population by unraveling some of the temporal structure as sustained power in a narrow high-frequency band (the gamma band). In this respect, the oscillatory activity adds information complementary to the information present in the firing rates. Our data suggest that parietal neurons synchronize their activity at gamma band frequencies to set up saccade plans, much alike neurons in visual cortex synchronize to facilitate the encoding of stimulus features (Fries et al., 2001; Gray and Singer, 1989; but see Thiele and Stoner, 2003). The present results also speak to the longstanding debate in the literature whether the posterior parietal cortex is more important for spatial attention or motor intention (Colby and Goldberg 1999; Andersen and Buneo 2002; Konen et al. 2007). In our study, we did not explicitly control spatial attention, only the movements of the eyes. Clearly, during the planning of an antisaccade, subjects will first attend to the stimulus and then shift and maintain their attention to the opposite location (Rizolatti et al. 1987). Hence, one could argue that the dynamic update of gamma-band activation that we found could simply represent a shift of spatial attention in the processing of antisaccades. How42
ever, an explanation by attentional processes alone falls short in accounting for why gamma-band power would increase in strength closer toward the execution of the saccade (see Figure 2.4). This observation seems more consistent with a motor planning effect. Our recent study on gamma band activity in parietal cortex during delayed double-step saccades provides also support for the motor intention explanation (Medendorp et al., 2007). In this study, we asked subjects to remember the locations of two sequentially flashed targets (each followed by a 2-s delay) and then make saccades to the two locations in sequence. Although both delay periods require sensory attention and memory, we found spatiallytuned sustained gamma band selectivity only in the second delay period, in relation to the intention to perform the saccades (Medendorp et al. 2007). Furthermore, along with the posterior parietal cortex, it is also widely accepted that the intention to make a saccade heavily involves the frontal eye fields (FEF) (see Munoz and Everling 2004 for review). Indeed, using intracranial recordings in patients, Lachaux et al. (2006) have shown increased gamma-band power during pro- and antisaccades in the human FEF. In the present study, we did not observe a reliable gamma band source in this region. This should however not be taken as evidence against the involvement of the FEF. It is known that null-findings in MEG could result from several factors. For example, field cancellation caused by differently oriented sources could attenuate frontal gamma activity. Thus, our results suggest that parietal gamma band synchronization reflects a mechanism to encode and amplify the saccade goals in the visuomotor system. What would be the further purpose of such dynamic memory fields related to motor intention? A possible explanation is that neurons that synchronize their activity, especially at higher frequencies, have a stronger effect on downstream areas (Tiesinga et al., 2004; Jensen et al., 2007). More specifically, in the oculomotor network, syn-
Gamma-Band Activity during Antisaccades
chrony may facilitate communication between parietal and frontal regions such as the frontal and supplementary eye fields. In light of our results, the gamma-band synchronization of neurons in the parietal fields may strengthen their projection to downstream regions, providing a putative mechanism to carry along unique spatial and motor information. In future work, it would be interesting to characterize the functional interactions between the
present parietal area and other areas within the oculomotor network, e.g., by means of a coherence analysis. Recent fMRI studies found that the type of representational codes that are being maintained in working memory bias frontal–parietal interactions (Curtis et al., 2005; Miller et al., 2005). Whether this effective connectivity in the oculomotor network is mediated mechanistically by gamma-band synchronization must await further studies.
43
2
44
Chapter 3 Neuronal synchronization in human parietal cortex during saccade planning
Adapted from Van Der Werf J, Buchholz VN, Jensen O, Medendorp WP (2009) Oscillatory activity in parietal cortex during saccade planning. Behaviour and Brain Research 205:329-335.
45
Chapter 3
A
3
major goal in neuroscience is to understand how the brain represents space and how this representation is used to generate behavior. For example, in order to act upon a visual stimulus, the brain must transform the spatial information from the visual system into a motor plan that ultimately leads to the motor commands for control of the necessary muscles. It is generally accepted that the posterior parietal cortex (PPC) plays a crucial role in this process of sensorimotor integration, being situated between higher-order visual areas and the cortical motor areas. To study the sensorimotor functions of the PPC, investigators often use the memory-guided response task. In this task, subjects have to remember the location of a brief peripheral cue for several seconds before a response is made toward it (Hikosaka and Wurtz, 1983). The strength of this task is that it separates temporally the sensory and motor responses from the delay-activity encoding the working memory in sensory-motor integration (Snyder et al., 1997; Medendorp et al., 2006; Brown et al., 2004; Bruce and Goldberg, 1985; Curtis et al., 2004). In monkey PPC, neurons in the lateral intraparietal area (LIP) have been associated with the generation of saccades (Barash et al., 1991; Gnadt and Andersen, 1988). Neurons in LIP have retinotopic receptive fields, and are arranged in a topographic fashion, with preferred tuning toward contralateral space (Blatt et al., 1990). During the memory interval of delayed saccades, these neurons show spatially-tuned elevations in their firing rate, even in the absence of visual input, which can be interpreted as the neural basis of working memory (Gnadt and Andersen, 1988). The nature of the working memory representation, however, has remained a point of debate. Some studies have associated the sustained delay activity with a retrospective stimulus representation while other reports have argued for a goal representation of the saccade (Andersen and Buneo, 2002; Colby and Goldberg, 1999; Got46
tlieb, 2002). It has also been shown that the activity in LIP is updated for changes of gaze, i.e., the brain re-computes and thus internally updates the working memory representations in eye-centered coordinates (Duhamel et al., 1992). In addition, neurons in LIP are characterized by gain fields, which describe changes in the amplitude of the response function depending on eye or head position (Andersen et al., 1985; Brotchie et al., 1995). With the development of functional magnetic resonance imaging (fMRI) methods, research over the past years has also advanced our understanding of the role of the human PPC in sensorimotor integration for saccades (Culham et al., 2006; Pierrot-Deseilligny et al., 2004; Curtis, 2006). This work has suggested a close correspondence of the human and monkey PPC. For example, human studies have reported topographically-organized maps of sustained activity during delayed-saccades (Sereno et al., 2001; Schluppeck et al., 2005, 2006; Medendorp et al., 2003, 2006; Jack et al., 2007). Furthermore, delay activity in human PPC exhibits internal spatial updating when gaze changes (Medendorp et al., 2003, 2008; Merriam et al., 2003, 2006) as well as gain field modulations (Desouze et al., 2000), analogous to what has been observed in macaque LIP. Notwithstanding these important findings, the higher-resolution temporal structure and spectral dynamics of these human parietal visuomotor processes have remained elusive because of the low temporal resolution of fMRI, which is the result of biological time constants (in order of seconds) involved in the hemodynamic response. This makes not only the interpretation of fMRI results in terms of the underlying neuronal activity difficult (Logothetis, 2008), but also compromises a direct relationship to intracranial findings from monkey and other primate studies. In the present paper, we review recent findings on the temporally fine-scaled dynamic aspects of sensorimotor integration in the human PPC, fo-
Neuronal synchronization in human parietal cortex during saccade planning
cusing on the role of neuronal synchronization in the coding and short-term storage of spatial representations in (pro-)saccade and antisaccade tasks. Although the main part of this review focuses on these mechanisms, we will also put these results in a broader perspective by relating them to work in monkeys as well as research with other measurement techniques (single-unit data, local field potentials, and BOLD-fMRI). With this approach, we will demonstrate that four major research techniques across species can be integrated into one framework, with the work on neuronal synchronization providing the link between humans and monkeys. For clarity, we emphasize that this article is not meant to provide an exhaustive, integrated view on the role of PPC in sensorimotor transformations at a broader scope, including all sensory modalities and effector systems.
Synchronized activity during visuomotor processing Invasively, rhythmic synchronous activity can be recorded by the local field potential (LFP), which represents the synchronized activity of a small group of neurons. At a larger scale, on the scalp, synchronized activity can be recorded using electroor magnetoencephalography (EEG/MEG), which allow measurements with sub-millisecond temporal resolution (Hari, 1999). Using spectral analysis methods, the power (i.e., synchronization) in various frequency bands can be studied as a function of time. Using this technique in combination with novel source-reconstruction methods, the spatial and temporal activation patterns employed by the PPC during various saccade tasks can be inferred. These results also allow for a direct link with related observations made in the monkey brain. Figure 3.1 guides the main work to be discussed, showing, on different time scales, the results of small-scale recordings from monkey area LIP (single cell recordings, Figure 3.1A), the local field potentials of this area (Figure 3.1B), related oscillations on the hu-
man scalp level (Figure 3.1C), as well as large scale BOLD measures (Figure 3.1D) of human parietal activation during the planning of pro- and antisaccades. While monkey and human PPC areas show increased firing rates or BOLD signals in delayed saccade tasks (Figure 3.1A and D, respectively), little is known about the neural mechanism that gives rise to the selective and sustained activity. Temporally correlated neuronal activity has been claimed to be important in coding memory representations (Tallon-Baudry et al., 1998; Howard et al., 2003; Jensen et al., 2002, 2007; Jokisch and Jensen, 2007). In a strict sense, this follows from the seminal work by Donald Hebb, now published about 60 years ago, suggesting that a memory representation can be encoded by two different processes: by reverberating activity through reciprocal connections in an assembly of neurons and by structural changes of these connections to enhance synaptic efficiency (Hebb, 1949). Expanding from this, high frequency (>30 Hz) oscillations could provide a mechanism to reverberate information (Tallon-Baudry et al., 1999; Pesaran et al., 2002) and control the flow of information in the brain (Salinas and Sejnowski, 2001). Regarding the latter, the impact of dendritic input on downstream neurons can be increased by high-frequency synchronization (i.e. gain modulation). If the downstream neurons also oscillate in synchrony, they are more receptive for their inputs (Fries, 2005). Given the local reciprocal connections between neurons, these mechanisms are optimally suited for the local maintenance of activity during a memory delay, forming transient assemblies of highly active neurons. Note that this does not exclude other possible functions of high-frequency synchronization, such as for example binding of perceptual input (Engel et al., 2001; Singer, 1999). Experimental evidence for these conceptual considerations of synchronization has only recently accumulated through improvement of measurement techniques and spectral analysis methods, 47
3
Chapter 3
3
such as multi-tapering (Percival and Walden, 1993). Gamma band synchronization (30-100Hz) is associated with active stimulus maintenance during short-term memory in both humans and monkeys (Jensen et al., 2002; Jokisch and Jensen, 2007; Pesaran et al., 2002; Tallon-Baudry and Bertrand, 1999; Scherberger et al., 2005; Kaiser and Lutzenberger, 2001). Also theta oscillations (4-8 Hz) have been claimed to be involved in short-term memory maintenance (Jensen and Tesche, 2002; Lee et al., 2005). It has been proposed that theta and gamma oscillations operate together to maintain multiple short-term memories, with the gamma oscillations related to the respective memories occurring at different phases of the theta rhythm (Lisman and Buzsaki, 2008; Lisman and Idiart, 1995). Although most of this evidence comes from hippocampal studies, there is increasing support for the hypothesis that such theta-gamma cross-frequency coupling serves as a more general mechanism in cortical functions (Canolty et al., 2006). Alpha band synchronization (8 -12 Hz) has been linked to the functional inhibition of areas disengaged in a certain task (Jokisch and Jensen, 2007; Klimesch et al., 2007; Medendorp et al., 2007; Sauseng et al., 2005; Rihs et al., 2007). Conversely, alpha band desynchronization has been related to increased processing or excitability of the respective areas (Sauseng et al., 2005; Romei et al., 2008; Thut et al., 2006; Yamagishi et al., 2005). In the present review, we will focus on some recent observations of synchronized activity during the preparation of saccadic eye movements. To date, only few studies have examined the role of synchronized activity during saccade tasks. One of the first studies was by performed by Okada and Salenius (Okada and Salenius, 1998). They instructed subjects to remember a target location for 3 s and then either execute a memory-guided saccade to that location or judge whether a new spatial cue appeared at that location or not. Their main focus was on alpha band activity during the retention 48
period, which was found to be present independent of preparation for the saccadic responses. The authors did not identify consistent sustained spectral power in the higher frequency range; perhaps a low signal-to-noise ratio prevented a demonstration of this. For example, local high-frequency enhancements are often surrounded by more widespread reductions (Lachaux et al., 2005; Shmuel et al., 2006), which may affect the resulting sensor-level amplitude differences such that it is hard to detect higher-frequency oscillations at a more global level, i.e. from the skull. Studies using intracranial recordings are typically less compromised by such effects, obtaining more focal activations (Pesaran et al., 2002; Lachaux et al., 2006). In the monkey, a recent study by Pesaran et al. (Pesaran et al., 2002) reported that during the delays in a memory-guided saccade tasks, there are broadband gamma oscillations (2590 Hz) in the LFP of area LIP, peaking at about 50 Hz. These authors found the LFP in the gamma band to be tuned to the direction of planned movements (see Figure 3.1B), changing its strength with behavioral state. The power of the high-frequency LFP became much larger when the animal planned a saccade, and rapidly decreased during the saccade execution. This spatially-tuned gamma band synchronization was interpreted as a mechanism for holding sustained neuronal activity, i.e., a neural basis for coding the spatial working memory for the saccade. Recently, similar broadband gamma oscillations have been found in the posterior parietal reach region in delayed-reaching tasks, in the range of 15 - 40 Hz (Scherberger et al., 2005; Buneo et al., 2003). These findings support the notion of the maintenance of a movement goal by high-frequency synchronization, but also raise new questions concerning the role of different frequency bands within the gamma range associated with different sensorimotor tasks. In this context, we should also emphasize that the gamma-band frequencies could vary considerably in individu-
Neuronal synchronization in human parietal cortex during saccade planning
3
Figure 3.1 Temporal dynamics of PPC activity during delayed prosaccades (left column) and antisaccades (right column) at increasing spatial and temporal scales. (A) single-cell recordings of monkey area LIP, Firing rate as a function of time (B) multi-cell recordings, or local field potentials (LFP) of monkey area LIP; power (color coded) as a function of time (C) magnetoencephalography (MEG) in human PPC, power (color coded) as a function of time, and (D) BOLD in human PPC. Plots represent differential activities between saccades in preferred (contralateral) and non-preferred (ipsilateral) directions as a function of time (in s). Although on different time axes, all scales show a similar pattern: an initial transient visual response for both pro- and antisaccades, with sustained activity in the same direction during prosaccades (left-hand panels) and in the opposite (motor) direction for antisaccades (right-hand panels).
als performing the same task (Hoogenboom et al., 2005). Taking individual gamma frequencies into account might improve analysis outcomes and facilitate interpretations across subjects and studies in functional terms. With the advent of better measurement tools and multi-tapering methods over the last years, as well as advanced spatial filtering techniques, it has become possible to detect gamma band activity at the skull level using MEG. Consistent with the observations in monkey area LIP, we have recently reported gamma band activity during sac-
cade planning in a region of the human posterior parietal cortex (Van Der Werf et al., 2008). In our experiment, subjects were cued with a brief visual stimulus in either the right or left visual field. Subjects had to remember the location of the stimulus and make a saccade toward it after a delay of 1.6 s. In response to the presentation of the stimulus, we found a broad-banded increase in gamma-band power (40-120 Hz), originating from occipital and posterior parietal regions in the contralateral hemisphere (see Figure 3.1C, left panel). Furthermore, this lateralized power enhancement was sustained 49
Chapter 3
3
in a more narrow frequency band (70-100 Hz) during the delay period, more focally in the posterior parietal cortex, where it increased in strength closer to the initiation of the saccade. Thus, human MEG and monkey LFP have revealed very consistent results regarding the directional selectivity of gamma band activity observed in posterior parietal cortex in the delayed-saccade task. In this respect, the subregion of the human PPC where the sustained activity was found may correspond to monkey area LIP, but that does not solve the question of what specific aspect of sensorimotor processing is encoded in the gamma band during the delay period. In more specific terms, one could ask whether the observed gamma band activity during the memory period relates to coding the remembered spatial sensory information (a retrospective sensory code) or whether it codes the motor goal of the planned saccade (a prospective motor code). Simple delayed saccades toward remembered visual stimuli cannot answer this question since these do not discriminate between the direction of the stimulus and direction of the saccade goal. One way to address this issue is using the antisaccade task, which separates sensory from motor goal representations by requiring participants to transform the location of a stimulus into an eye movement to the opposite visual field (Munoz and Everling, 2004; Hallett, 1978). While a number of studies have reported that the parietal cortex is engaged in stimulus-to-goal mapping during antisaccades (see Figure 3.1, right panel) (Zhang and Barash, 2000, 2004; Everling et al., 1998; Moon et al., 2007; Medendorp et al., 2005; Nyffeler et al., 2008), the effects on spectral power in relation to the neural dynamics of this process were only recently described by the Van Der Werf et al. study (Van Der Werf et al., 2008). To test whether the observed gamma band relates to the sensory stimulus or to the motor goal representation, we exploited a memory-guided an50
tisaccade task in MEG (Van Der Werf et al., 2008). We reasoned that in case of motor goal coding, the directional selectivity of gamma band activity should reverse after presentation of the stimulus (Zhang and Barash, 2000; Medendorp et al., 2005). However, if the directional-selective gamma activity in parietal cortex relates to a visual memory representation, it should not remap during the delay of a planned antisaccade. The results are shown in Figure 3.1C, right panel, which clearly indicate that the temporal structure of human parietal gammaband activity reflects the process of target remapping for antisaccades. As shown, after presentation of the stimulus, there is first the broad-band gamma response in the contralateral hemisphere. However, the stimulus-related bias quickly disappears during the delay interval and turns into a bias in the ipsilateral hemisphere to the stimulus, which is the hemisphere contralateral to the direction of the saccade goal. This stimulus-to-goal mapping suggests that the sustained gamma-band synchronization in human parietal cortex represents the planned direction of the saccade, not the memorized stimulus location. It should be noted that the parietal source that underlies the sustained gamma response during antisaccades overlapped with the neural locus coding the sustained response during the saccade planned toward the stimulus (not shown). These human results mimic, on different time scales, the results of corresponding experiments in BOLD-fMRI (Figure 3.1D). Indeed recent studies, simultaneously recording electrophysiological and hemodynamic (BOLD/PET) signals, have shown high positive correlations between increases in the hemodynamic signals and spectral power in the gamma band, in spatially overlapping networks (Lachaux et al., 2007; Logothetiset al., 2001; Niessing et al., 2005; Nishida et al., 2008). The MEG results also bridge the gap to the monkey work of Zhang and Barash (Zhang and Barash, 2000, 2004), who demonstrated, using delayed an-
Neuronal synchronization in human parietal cortex during saccade planning
tisaccades, that a stimulus direction is remapped in the firing rates of LIP neurons in order to code the direction for a saccade (see Figure 3.1A, right panel). Of all the neurons that remapped, Zhang and Barash reported an average inversion time of about 400 ms, which is also relatively close to the average inversion time of 500 ms found in the human. Note that the inversion time may depend on context information, i.e., whether or not the type of response (pro/anti) is known prior to stimulus presentation, which might explain differences with other studies (Gail and Andersen, 2006). Regarding the visuomotor updating, not all neurons in the Zhang and Barash (Zhang and Barash, 2004) study showed the remapping effect: some neurons coded consistently the location of the stimulus, while others exclusively encoded the movement goal. Thus, LIP activity as studied by multi-unit firing rates reflects all aspects of sensorimotor control: the sensory stimulus, the motor output, as well as the transformation between these representations. Does the oscillatory activity also reflect all aspects of sensorimotor control? To address this issue in the human, we decomposed the gamma band activity in stimulus and goal components under the assumption that saccades toward the memorized location and away from it have the same stimulus component but an opposite goal component. As a result, we found a broad-band, but transient, stimulus-induced component that did not persist into the memory period. In contrast, the goal component showed a sustained component during the entire delay interval, increasing in strength toward the execution of the saccade. Thus, the goal for a saccade is ultimately represented in the parietal gamma band activity and not the memory of the visual stimulus. Interestingly, the activity in the lower frequency bands, i.e. the alpha band (8-12 Hz), revealed quite a different picture. After stimulus presentation, power in the alpha band increased in the ipsilateral hemisphere, consistent with the interpretation of
functional disengagement/inhibition. Also during the delay period (0.6 – 1.6 s.), the alpha band lateralization was stronger linked to the direction of the initial stimulus than to the saccade goal (Van Der Werf et al., 2008). To reconcile this observation of neuronal synchronization with the firing rate findings by Zhang and Barash, these results imply that only those neurons that represent the goal of the saccade show a sustained synchronization at gamma band frequencies, and neurons that are involved in representing the stimulus position do not. If true, this would also be consistent with the hypothesis that high-frequency synchronous firing of a group of neurons exerts a stronger effect on downstream areas (Jensen et al., 2007; Tiesinga et al., 2004). One further clue, in this respect, may come from a recent study by our group in which we investigated gamma band activity during double-step saccade planning (Medendorp et al., 2007). Subjects remembered the locations of two sequentially flashed targets (each followed by a 2-s delay), presented in either the left or right visual hemifield, and then made saccades to the two locations in sequence. In this study, we found directional-selective gamma-band synchronization only to occur at some point during the second delay period, in a 1-s period prior to the execution of the saccades. Again, such synchronization should not be interpreted as a retrospective stimulus representation but rather as a neuronal correlate for preparatory set, coding goal information for the upcoming saccade. Experiments that explicitly address this issue in a spatial updating task are addressed in the next chapter (chapter 4) and have appeared in abstract form (Buchholz et al., 2008). Admittedly, the results of all these studies should be interpreted with caution (Jerbi et al., 2008). For example, the integrative effect of field potentials over a large number of neurons that code the goal may obscure the synchronized tuning to the visual stimulus of a smaller fraction of 51
3
Chapter 3 neurons. Another point of remark is that subjects, who plan a saccade away from a visual stimulus, also shift their attention away from that stimulus. Perhaps, gamma synchronization reflects activity of a system responsible for both attentional tuning and motor planning of saccadic eye movements (Reva and Aftanas, 2004). Further work is needed to investigate whether these processes operate in the same frequency bands, or rather independently at the neural level (Wyart and Tallon-Baudry, 2008; Kayser and Konig, 2004).
Functional implications
3
Both the human and monkey experiments provide converging evidence that parietal gamma-band synchronization plays a role in saccade planning. One interpretation of this synchronization relates to the coding of the goal of the saccade - a working memory for movement planning - which finds clear support in the studies reviewed. Other studies have related gamma band activity to more efficient processing of sensory changes (Kaiser et al., 2006; Womelsdorf et al., 2007), accuracy in sensory tasks requiring behavioral responses (Kaiser et al., 2007, 2008) and increased response speed in sensory-motor task (Gonzales Andino et al., 2005; Womelsdorf et al., 2006). These findings suggest that synchronization at the gamma band frequency increases efficiency in sensory-motor integration networks and most probably plays an important role in memory-guided movement planning. On a neurophysiological level, neuronal synchronization at frequencies higher than 30 Hz has been suggested to increase both inter and intraareal communication: when a local assembly of neurons fires coherently, this leads to non-linear increases in input gain in a downstream neuron (Salinas and Sejnowski, 2001). If several inputs are provided simultaneously to a receiving neuron, synaptic integration can induce a rapid change in membrane depolarization, thereby increasing the probability of a spike. Furthermore, if the down52
stream assembly of neurons fires in coherence with the source neurons, the input arrives at times optimal for reception, thereby increasing the gain even further (Womelsdorf et al., 2007;Zeitler et al., 2008). In humans, regions on the dorsal visual pathway have indeed been shown to oscillate coherently in a spatial attention task (Siegel et al., 2008). Although not conclusive, this is in line with the interpretation that parietal cortex increases its gain on downstream motor areas at the moment it provides a veridical goal representation for a subsequent saccade. Besides a role in gain control, oscillations have also been suggested to provide a temporal framework in which the more excited neurons fire earlier in the oscillatory cycle (Fries et al., 2007; Kayser et al., 2009). This recoding of spikes into phase-values allows for fast prioritizing and read-out of information that is already implicitly present in the firing rates. Suggestive of a role for spike-field coherency in visuomotor transformations are the recent findings that during delayed sensorimotor tasks PPC neurons fire preferentially in a particular phase of a local or distant oscillatory rhythm (Pesaran et al., 2002, 2008; Scherberger et al., 2005). Of note is the temporal distinction between local and distant spike-field coherence. When spikes were temporally entrained to a local oscillation for the entire delay period, entrainment to a distant oscillation was observed only right after goal presentation and just prior to saccade execution. For long-range communication, this would provide an ideal mechanism for read-out of information by downstream areas. It could be argued that this transfer of information between cortical sites is only needed when an input (a spatial stimulus) enters the brain, or an output (movement) leaves the brain. Hence, inter-areal spike-field coherence was observed only at the beginning and end of the memory-guided movement trials (Pesaran et al., 2008). Conversely, intra-areal spike-field coherence was observed throughout the delay period (Pesaran et al., 2002; Scherberger et
Neuronal synchronization in human parietal cortex during saccade planning
al., 2005). Since this temporal coding of spikes may also facilitate neuronal communication between the different cortical layers of the same area, these findings might reflect a mechanism to maintain a goal representation in working memory. These ideas might help to integrate the findings from the Van der Werf et al. study (Van Der Werf et al., 2008), with spiking results from the Zhang and Barash study (Zhang and Barash, 2000, 2004). Where Van der Werf et al. found the sustained gamma band to code only for the saccade goal, Zhang and Barash found LIP neurons to code predominantly, but not exclusively, for the saccade goal. During delayed antisaccades, a local synchronization in the gamma band frequency might represent a mechanism to prioritize the spikes representing the goal of the saccade, whereas firing rates in general may also reflect stimulus information necessary to perform the transformation between sensory and motor representations. The low frequency oscillations, in contrast, may control neuronal communication by allocation of neural resources, and hence, allow an effective information transmission. In sum, the results presented here shed more light on the role of high frequency neuronal synchronization in parietal cortex during human
saccade planning. We have suggested that this synchronization relates to the coding of the goal of the saccade, and is updated when the direction of the saccade is changed. We have further discussed how these findings provide an important link between invasive data reported in monkey studies and previous fMRI studies in humans. In future studies, it remains to be established how well these parietal findings generalize to other types of movements, e.g. planning reaching and grasping movements (Culham et al., 2006; Gail and Andersen, 2006; Galletti et al., 2003). Novel methodology will further help to build an integrated view on the role of the PPC in sensorimotor integration. For example, a future promise for humans investigations are invasive EEG measurements (Lachaux et al., 2006; Jerbi et al., 2008), which have very high signalto-noise ratio, no uncertainty related to source reconstruction, and are not contaminated by the various artifacts and confounding signals (Jerbi et al., 2009; Yuval-Greenberg et al., 2008) in scalp recordings. These and other studies will not only provide further basic insights in parietal sensorimotor mechanisms, but hopefully also lead to a better understanding and new treatments of the disorders and diseases that affect them.
53
3
54
Chapter 4 Reorganization of oscillatory activity in human parietal cortex during spatial updating
55
Chapter 4
T
4
he posterior parietal cortex (PPC) plays an important role in the transformation of spatial representations from perception to action. In particular, activity in the lateral intraparietal area (LIP) of monkey PPC and homologous areas in the human PPC has been associated with specialized spatial functions, including the control of spatial attention (Silver and Kastner, 2009; Bisley and Goldberg, 2010; Liu et al., 2010), working memory (Pesaran et al., 2002; Curtis et al., 2004) and saccade planning (Sereno et al., 2001; Andersen and Buneo, 2002; Zhang and Barash, 2004; Liu et al., 2010). The neural architecture of LIP is characterized by primarily eye-based, gaze-centered response fields (Andersen and Buneo, 2002; Patel et al., 2010). Moreover, within the gaze-centered neuronal population of LIP, activity has been demonstrated to remap in order to compensate for intervening saccades (Duhamel et al., 1992; Colby et al., 1995; Medendorp et al., 2003; Merriam et al., 2003). Gain field modulations, the scaling of neuronal firing rates by eye- and head-position, have been suggested to implicitly transform these spatial representations into other gaze-independent (e.g. head- or body-centered) reference frames (Andersen et al., 1985; Chang et al., 2009). Despite these important insights, little is known as to how the respective neurons cooperate during exchange of information within and across reference frames. A prime mechanism for cooperation is rhythmic neuronal synchronization, which comes about in various frequency bands. In general, gamma band oscillations (> 30 Hz) have been implicated in local processing (Fries, 2009), while alpha band oscillations (8-12 Hz) reflect functional inhibition (Klimesch et al., 2007). Indeed, intracranial local field potential recordings have shown that neurons in monkey area LIP synchronize their activity in a direction-selective fashion during the coding of a working memory for a saccade (Pesaran et al., 2002). Corresponding observations have recently also been made in humans, using magnetoenceph56
alography (MEG) (Medendorp et al., 2007; Van Der Werf et al., 2008, 2009, 2010). But without varying eye position, the question remains unanswered whether these parietal oscillations are related to the construction of a gaze-independent spatial representation, or are a manifestation of the saccade goal, encoded in gaze-centered coordinates. To discriminate between these two possibilities, we applied MEG to record oscillatory brain activity from human subjects while they produced intervening saccades between viewing a goal target and generating an eye movement toward its remembered location. While the target remained stable in gaze-independent coordinates (i.e., relative to head/body), its remembered location must be updated to compensate for the intervening saccade in gaze-centered coordinates. Here we demonstrate, by exploiting the direction selectivity of the power in the various frequency bands, a reorganization of oscillatory activity during spatial updating. Parietal gamma band synchronization represents and updates the goal direction of a saccade in a gaze-centered reference frame. Power in the alpha band reflects a regulatory mechanism in spatial updating, inhibiting the retrospective target representation and facilitating the updated target representation for the saccade.
Methods Participants Twenty-two naïve participants (7 female/15 male; mean age 26.5 years), free of any neurological or psychiatric disorders, volunteered to participate in the study. All participants provided written consent according to guidelines of the local ethics committee (CMO Committee on Research Involving Humans subjects, region Arnhem-Nijmegen, the Netherlands). MEG recordings Participants sat upright in the MEG system, viewing a stimulus screen that was positioned 40 cm
Reorganization of oscillatory activity in human parietal cortex during spatial updating in front of them. Stimuli were generated with Presentation 9.10 software (Neurobehaviroal Systems Inc. Albany). Using an LCD video projector (SANYO PLC-XP41, 60 Hz refresh rate), these stimuli were projected onto the screen via two front-silvered mirrors. MEG data were recorded continuously using a whole head system with 275 axial gradiometers (CTF Systems Inc., Port Coquitlam, Canada). Head position with respect to the sensor array was continuously measured using localization coils fixed at anatomical landmarks (the nasion and at
the left and right ear canal). Horizontal and vertical electrooculograms (EOG) were recorded using electrodes placed below and above the left eye and at the bilateral outer canthi. Impedance of all electrodes was kept below 5 kΩ. During the experiment, the EOG recordings were continuously inspected to ensure that participants were vigilant and correctly performing the task. Furthermore, the electrocardiogram (ECG) was recorded with electrodes (impedance < 50 kΩ) attached above the right clavicula and under the last false rib on the
4
Figure 4.1 Experimental design. A. Trial timing. Each trial started with a baseline period of 1 s., during which subjects were instructed to fixate a centrally presented fixation cross. Next, a stimulus was presented for 0.1 s, randomly right or left of central fixation, followed by a 2 s delay period (1st delay). The 2nd delay started with the relocation of the fixation cross unpredictably to the left or right of the remembered target location, shortly followed by the subject’s gaze. Due to the intervening saccade, the remembered stimulus location remained either in the same, or shifted to the opposite visual hemifield. After another 2 s. delay, the new fixation cross disappeared, signaling the subject to make an eye movement to the remembered target location. Reappearance of the central fixation cross signaled the start of a new trial. B. Four different trial types could be distinguished, depending on the remembered location of the target relative to gaze before and after the intervening saccade. Corresponding EOG traces for each trial type are presented at the bottom, grouped together based on the position of the flashed stimulus. Schematic representations of the two possible refixation positions (colored), the stimulus position (circle) and the central fixation cross (white cross) are presented in the right-hand panels, with the colors of the refixation crosses matching the color of the associated EOG trace.
57
Chapter 4 left side. All signals were low-pass filtered at 300 Hz, sampled at 1200 Hz, and then saved to disk. For each participant, a full brain anatomical MR image was acquired using a standard inversion prepared 3D T1-weighted scan sequence (FA=15º; voxel size: 1.0 mm in-plane, 256 x 256, 164 slices, TR=0.76 s; TE=5.3 ms). A 1.5 T whole-body scanner (Siemens, Erlangen, Germany) was used to record the anatomical MRIs, with reference markers at the same locations as during the MEG recording, to allow alignment of the individual MEG and MRI data in later analyses.
4
Experimental paradigm Subjects performed an intervening saccade task, shown in Figure 4.1. Each trial began with the subject fixating centrally at a small white cross, presented on the screen. After a baseline period of 1.5 s, a target stimulus was flashed for 100 ms in the left or right visual hemifield, horizontally at a mean eccentricity of 3º or 9º, and positioned vertically at a polar angle < 45º relative to the horizontal meridian. Targets were jittered slightly in eccentricity (2° visual angle) to make them less predictable. After a 2 s delay period (the 1st delay period), the fixation
Trialtype
Screen positions: Target (T), Gaze refixation (G) -15
-9
-3
1
3
9
G
T
2
T
3
G
4 G
6 8
T T
5 7
cross jumped to a new position, at a horizontal eccentricity of 3º, 9º, or 15º (jitter 2º), in either the left or right visual field, unpredictable to the subject. Subjects were instructed to immediately saccade to the new fixation position (i.e. the intervening saccade), which was presented for a duration of 2 s (2nd delay period). The offset of the fixation cross signaled the subjects to look at the remembered location of target. The refixation positions were chosen such that the desired amplitude of this saccade was on average 6º (jittered in the interval 4º - 8º). Subsequently, 0.7 s later, the central fixation cross reappeared, indicating the start of a new trial. Trials were presented in 20 blocks of 30 trials each, with the different blocks separated by a brief self-paced resting period. Essentially, the paradigm had four different conditions regarding the remembered location of the target relative to gaze before and after the intervening saccade. It either remained to the right (RR condition), or remained to the left (LL), or it shifted from right to left (RL), or moved from left to right (LR). In contrast, during the interveningsaccade task, the location of the target is invariant in a gaze-independent coding frame, such as a
T T
G
T T
G
G
G
15 G
Target re. to gaze (Left (L)/Right (R)) 1st delay
2nd delay
Gaze
R
R
R
R
L
R
R
R
L
R
L
R
L
R
L
L
L
R
L
R
L
L
L
L
Table 1 Schematic representation of the 8 conditions. Three components were manipulated: Stimulus location (left or right, at ~3o or ~9o), final saccade direction (left or right of gaze, amplitude always ~6o), and gaze direction (left or right of central fixation, at 3o, 9o or 15o).
58
Reorganization of oscillatory activity in human parietal cortex during spatial updating head, body or world-fixed frame. Behavioral analysis Electrooculographic (EOG) data from each subject were inspected online to ensure high vigilance levels and correct performance of the task. Figure 4.1B shows the EOG traces (horizontal component) of a typical subject during 20 trials of each of the four conditions. A schematic representation of the respective target (o) and fixation position (+) is flanked on the right-hand side. For each condition, the subject keeps stable fixation during the 1st delay period. The subsequent intervening saccade brings the eyes’ fixation point either to the left or to the right of the remembered target location, which is subsequently well maintained till the end of the 2nd delay period. The saccade to the remembered target location is made after the fixation spot was turned off. Note that, for every remembered target position, there are opposing eye movement vectors with about equal amplitudes. The EOG recordings in all 22 subjects confirmed that they followed the instructions correctly in most trials. Trials in which subjects broke fixation, blinked or performed incorrectly otherwise, were excluded from further analysis. There was no significant difference in the number of rejected trials among the 4 task conditions, specified above (one-way ANOVA, F(3,63) = 2.4; p>0.05). On average 398 +/- 89 (SD) trials per participant were accepted for further analysis. Also, reaction times for the memory-guided saccades times (LL: 181 ms; RR: 185 ms; LR: 187 ms; RL: 182 ms) did not differ between the four conditions (one-way ANOVA, F(3,63) = 2.0, P > 0.05). MEG data analysis Data were analyzed using Fieldtrip software (http://www.ru.nl/neuroimaging/fieldtrip), an open source Matlab toolbox for neurophysiological data analysis developed at the Donders Institute for Brain, Cognition and Behaviour. From the trials that survived the exclusion criteria described
above, data segments that contained muscle activity or jump artifacts in the SQUIDS were excluded using semi-automatic artifact rejection routines. Furthermore, independent component analysis (ICA) was used to clean the sensor-level data of cardio-magnetic artifacts as well as eye-muscle artifacts due to the eccentric fixation during the 2nd delay period in our paradigm. More specifically, we excluded the components that correlated highest (r>0.15) with either the EOG or ECG signal and had a spatial topography associated with ocular or cardiac magnetic effects (Barbati et al., 2004). Also components whose effects were topographically located around the eyes were excluded from the data. For the sensor level analysis, an estimate of the planar gradient was calculated for each sensor (Bastiaansen and Knosche, 2000). The horizontal and vertical components of the planar gradients estimated using the signals from the neighboring sensors approximate the signal measured by MEG systems with planar gradiometers. The planar field gradient simplifies the interpretation of the sensor-level data since the maximal signal is located above the source (Hamalainen et al., 1993). Power spectra were computed separately for the horizontal and vertical planar gradients of the MEG field at each sensor location and the sum of both was computed to obtain the power at each sensor location irrespective of the orientation of the gradient. Time-frequency representations (TFR), estimating the time course in power, were computed using a Fourier approach, applying a sliding tapered window, with neighboring time-points temporally segregated by 0.05 s. Because the gamma band is typically much wider and therefore better characterized with more spectral concentration (Hoogenboom et al., 2006), we analyzed two frequency ranges separately. For the lower frequency band (5-40 Hz), we used sliding windows of 0.5 s and a Hanning taper. This resulted in a spectral smoothing of roughly 3 Hz. For the higher frequency band (40-130 Hz) we applied a multi-taper approach 59
4
Chapter 4
4
(Percival and Walden, 1993) using a sliding window of 0.4 s and 11 orthogonal Slepian tapers. This resulted in a spectral smoothing of approximately 14 Hz. To localize the neural sources of the various spectral components, we applied an adaptive spatial filtering (or beamforming) technique (Dynamic Imaging of Coherent Sources, DICS) (Gross et al., 2001; Liljestrom et al., 2005). First, we divided a template brain voIume (international Consortium for Brain Mapping template; Montreal Neurological Institute, Montreal, Canada) into a regular 1 cm three-dimensional grid. We then warped each subject’s MRI to fit this template MRI and the template’s grid, after which we warped the grid back to fit the subject’s original MRI to obtain a grid in MNI coordinates for each subject. This procedure allowed us to directly compare grid points across subjects in MNI-space without the need to normalize. For each subject and for each grid point, a spatial filter was constructed that passes activity from this location with unit gain, while attenuating activity from other locations (Gross et al., 2001). This filter was computed from forward models with respect to dipolar sources at each grid point (the lead field matrix) and the cross spectral density between all combinations of sensors at the frequency of interest. We used realistic single-sphere head models from each subject’s individual MRI to calculate the lead field matrix (Nolte et al., 2003). For every single subject, the source power was estimated to the same baseline interval that was used for the sensor-level analysis. Statistical inferences We computed the task-related changes in power in various frequency bands relative to average power in the baseline period (see Figure 4.1). The highfrequency baseline power was computed over the period from -0.4 to -0.2 prior to the presentation of the stimulus, using a 0.4 s wide sliding window. The low-frequency baseline was determined across the 60
interval -0.35 to -0.25 s, using a 0.5 s sliding timewindow. Thus, in effect, the baseline period was equal for both frequency ranges: -0.6 to 0 s prior to stimulus onset. We expressed the difference in log power between the respective delay periods and the baseline as a t-score for each subject and for each condition. The resulting t-scores were transformed into z-scores (Bauer et al., 2006; Medendorp et al., 2007) to obtain normalized estimates of power differences. The resulting z-scores, which are well normalized for intrasubject variance, were pooled across subjects (zgroup = 1/√N ∑zi with zi being the z score of the i-th subject). In the 1st delay, statistical significance of the power modulations was tested at the sensor level by using a non-parametric clustering procedure (Nichols and Holmes, 2002; Maris and Oostenveld, 2007). In this procedure, cluster-level test statistic are defined by pooling the z-scores of neighboring sensors showing the same effect in a given timefrequency window of interest. In a nonparametric statistical test, the type-I error rate is controlled by evaluating the cluster-level test statistic under the randomization null distribution of the maximum cluster-level test statistic. In our analysis, this was obtained by randomly permuting the data between two conditions within every participant. By creating a reference distribution from 1000 random sets of permutations, the p-value was estimated as the proportion of the elements in the randomization null distribution exceeding the observed maximum cluster-level test statistic. The significant channels were used for further analysis of the power changes during the 2nd delay, i.e. after the intervening saccade, using the randomization approach. A nonparametric approach was also applied to test for statistical significance at the source level, clustering together neighboring voxels exhibiting a similar effect in a predefined volume of interest, comprising of the occipital and parietal cortices. No reliable frontal sources were observed.
Reorganization of oscillatory activity in human parietal cortex during spatial updating Isolating task-dependent spectral power The paradigm was designed with 8 different types of trials. Table 1 provides an overview of the screen positions (horizontal components) of the target position (T) and the gaze refixation location (G), with the initial central fixation at 0º. Recall that in the actual task the respective positions were jittered slightly. The 8 different trial types can be divided into RR, RL, LL, and LR conditions, as described above. In our analyses, we exploited the hemifieldspecific lateralization of power to compare conditions in which the target shifts sides relative to gaze (RL, LR) versus conditions in which the target remains at the same side from gaze (RR, LL). To assess the hemifield-specific lateralization of power during the 1st delay, we organized our trials into two groups, irrespective of target eccentricity. We compared trials with the target stimulus presented in the right visual field (RR and RL trials) to those with a target presented in the left visual field (LL and LR trials). To evaluate these data in a pooled comparison across hemispheres, we combined the hemifield-specific changes in power in terms of contra- versus ipsilateral target locations. Lateralized power during the 2nd delay interval may reflect the contributions of a number of factors: the updated target location in gaze-centered coordinates (i.e., the direction of the planned saccade), the target memory in gaze-independent (head/body) coordinates, the direction of gaze fixation (i.e., eye position), or a combination hereof. We isolated either of these factors by analyzing separately subsets of trials in which the other two factors remained invariant in hemifield. In other words, to analyze the hemispheric laterality of power in relation to gaze-dependent target updating (Gd), we compared trials in which the head-centered hemifield of the target and the hemifield of fixation remained the same for both delays, but the gaze-centered hemifield of the remembered target changed. We computed the directional power selectivity for both the gaze-indepen-
dent target representation (Gi) and the direction of eye fixation (Ge) in a similar manner. In terms of contra and ipsilateral locations, we computed for sensors overlying the right hemisphere: Gd = (P4 – P1) + (P8 – P5), Gi = (P6 – P4) + (P5 – P3), and E = (P8 – P6) + (P3 – P1), in which Pi represents the power during the 2nd delay in trial type i in Table 1. For the corresponding calculations for the sensors covering the left hemisphere, we only flipped the sign in the above equations. The isolated contributions were pooled across hemispheres, resulting in the combined hemisphere-specific changes in power during the 2nd delay period of our paradigm.
Results We examined the role of neuronal synchronization in the representation of memorized visual targets across intervening eye movements. In our test, subjects fixated at a central point while a target was briefly cued into the retinal periphery. After a delay, subjects switched fixation points (the intervening saccade), and then after another delay, looked to the remembered location of the target. On half of the trials, the intervening saccade made the remembered location of target change sides relative to the gaze line, while on the other half of the trials, the remembered target stayed on the same side relative to gaze. We assessed the laterality of spectral power during the two delay intervals to characterize the oscillatory activity in terms of gazedependent target updating (Gd), gaze-independent target coding (Gi), and gaze direction (Ge). Contralateral gamma band activity after target presentation We start the description of our results with a focus on the high-frequency power modulations (> 40 Hz) during the 1st delay interval. During this interval subjects have to memorize a location of a target that served as a goal only after an intervening eye movement, which could be directed either leftward or rightward. Figure 4.2A shows the scalp topog61
4
Chapter 4 raphy of 40-60 Hz gamma band activity, averaged across subjects, during 4 consecutive non-overlapping time intervals, each covering 500 ms of the 1st delay interval. Regions with warmer (red) colors (positive z-scores) show a preference for contralateral targets; regions with cooler (blue) color (negative z-scores) have a bias for ipsilateral targets. We marked the positions of sensors that showed a significant effect in a pooled analysis across hemispheres (P
